Asner Et Al 2004 Grazing Systems, Ecosystem Responses, and PDF

16 Oct 2004 10:42 AR AR227-EG29-08.tex AR227-EG29-08.
sgm LaTeX2e(2002/01/18) P1: GCE

10.1146/annurev.energy.29.062403.102142
Annu. Rev. Environ. Resour. 2004. 29:261–99

doi: 10.1146/annurev.energy.29.062403.102142
Copyright c 2004 by Annual Reviews. All rights reserved
First published online as a Review in Advance on July 26, 2004
GRAZING SYSTEMS, ECOSYSTEM RESPONSES, AND

GLOBAL CHANGE
Gregory P. Asner,1,2 Andrew J. Elmore,1 Lydia P. Olander,1
Annu. Rev. Environ. Resourc. 2004.29:261-299. Downloaded from arjournals.annualreviews.org
Roberta E. Martin,1 and A. Thomas Harris1

1
Department of Global Ecology, Carnegie Institution of Washington, Stanford,
California 94305; email: aelmore@globalecology.stanford.edu,
lolander@globalecology.stanford.edu, robin@globalecology.stanford.edu,
by Technische U. Muenchen on 05/03/05. For personal use only.
thomas@globalecology.stanford.edu
2
Department of Geological and Environmental Sciences, Stanford University, Stanford,
California 94305; email: gasner@globalecology.stanford.edu
Key Words agriculture, deforestation, desertification, land-use change, woody

encroachment
■ Abstract Managed grazing covers more than 25% of the global land surface and
has a larger geographic extent than any other form of land use. Grazing systems per-
sist under marginal bioclimatic and edaphic conditions of different biomes, leading to
the emergence of three regional syndromes inherent to global grazing: desertification,
woody encroachment, and deforestation. These syndromes have widespread but differ-
ential effects on the structure, biogeochemistry, hydrology, and biosphere-atmosphere
exchange of grazed ecosystems. In combination, these three syndromes represent a
major component of global environmental change.
CONTENTS
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
THE GLOBAL FOOTPRINT OF GRAZING SYSTEMS . . . . . . . . . . . . . . . . . . . . . . 263
Basic Demographic Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
GIS Analysis of Bioclimatic and Edaphic Conditions . . . . . . . . . . . . . . . . . . . . . . . 263
Regional Syndromes in the Global Grazing Footprint . . . . . . . . . . . . . . . . . . . . . . . 269
DESERTIFICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
Ecosystem Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
Biogeochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
Trace-Gas Emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
Hydrology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
Climate Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
WOODY ENCROACHMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
Trace-Gas Emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
1543-5938/04/1121-0261$14.00 261
16 Oct 2004 10:42 AR AR227-EG29-08.tex AR227-EG29-08.sgm LaTeX2e(2002/01/18) P1: GCE
262 ASNER ET AL.
Hydrology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
DEFORESTATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Trace-Gas and Aerosol Emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
Hydrology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
ECOSYSTEM RESPONSES TO MANAGED GRAZING . . . . . . . . . . . . . . . . . . . 286
Response Typologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
Knowledge Gaps and Research Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
INTRODUCTION
Managed grazing occupies more than 33 million square kilometers or 25% of the
global land surface, making it the single most extensive form of land use on the
planet. Managed grazing systems are defined here as any geographically extensive
operation designed for the production of animals for consumption, including for
meat, milk, and any major animal products. Recent work indicates that managed
grazing systems have increased more than 600% in geographic extent (from about
5.3 M km2) during the past three centuries (1). More than 1.5 billion “animal units”
(AU)1 were present in managed grazing systems on Earth in 1990 (2).
Despite these impressive statistics, there are surprisingly few synthetic reports
on how managed grazing systems affect global ecological, atmospheric, or hydro-
logical processes. The predominance of site-specific perspectives on ecosystem
responses to managed grazing has led to a fragmented understanding of this im-
portant land use as a contributor to global environmental change. Given the im-
portance of managed grazing systems to the subsistence of the human population,
and given the increasing role that land degradation plays in determining the long-
term sustainability of pastoral practices throughout the world (3), a global-scale
overview based on available scientific data is overdue.
In this review, we develop a perspective on ecosystem responses to managed
grazing. We employ a basic geographic information system (GIS) analysis and a
literature review to determine the environmental “footprint” of grazing systems
throughout the world. In this case, the footprint represents responses of ecosys-
tems to managed grazing relative to global bioclimatic and edaphic variability. We
find that managed grazing occupies bioclimatically and edaphically marginal lands
throughout much of the world and that these conditions predispose current range-
lands to three regional syndromes—desertification, woody encroachment, and de-
forestation. We use these syndromes as an organizing framework to synthesize
1
An animal unit (AU) is defined as the number of cattle, buffalo, sheep, goats, horses, and
camels weighted by their relative size and growth rates [AU = n (cows + buffalo) + 0.2
n (sheep + goats) + 1.2 n (horses + camels)] (2).
GRAZING SYSTEMS AND GLOBAL CHANGE 263
the impacts of managed grazing on ecosystem structure, biogeochemistry, hydrol-

ogy, and biosphere-atmosphere interactions. We contend that these syndromes,
when taken in combination, represent a major component of global environmental
change.
THE GLOBAL FOOTPRINT OF GRAZING SYSTEMS

Basic Demographic Patterns
Combining global grazing area in 1990 (1) with country-level stocking rates [AU
per grazed area; (2)], we estimated the geographic extent and intensity of managed
grazing systems worldwide (Figure 1). The five countries with most land area in
grazing systems are Australia, 4.4 M km2; China, 4.0 M km2; United States, 2.4
M km2; Brazil, 1.7 M km2; and Argentina, 1.4 M km2. However, based on the
fraction of total land area each nation uses for grazing, Mongolia, Botswana, and
Uruguay lead with 80%, 76%, and 76%, respectively. Countries with the highest
stocking rates are Malaysia, 320 AU km−2; India, 272 AU km−2; N. Korea, 213 AU
km−2; and Vietnam, 184; others are found in central Europe and the Middle East.
Countries containing large tracts of dryland grazing systems, such as in Australia,
Argentina, and the United States, have low stocking rates.
Our GIS analyses of grazing extent, grazing intensity, and human demographic
statistics reveal very few correlates (Figure 2). Grazing land area is well correlated
with total land area of each country (r = 0.79, p < 0.05). However, the total number
of animal units per country is weakly correlated with both grazing area (r = 0.50,
p < 0.05) and total land area (r = 0.60, p < 0.05). Grazing intensity—stocking
rates at the country level—is not correlated with grazing area or total land area
(Figure 2). By far the strongest relationship is found between human population
and the number of grazing animals per country (r = 0.91, p < 0.01). Human
population growth is likely to increase demand for meat and dairy products, which
will have to be met by a combination of increasing intensification and continued
extensification, as has been observed over the past 300 or more years. Evidence
reported in the following sections suggests that grazing extensification cannot
occur without continued major changes in global land cover, and that intensification
will also have significant environmental impacts.
GIS Analysis of Bioclimatic and Edaphic Conditions

The global footprint of managed grazing, which implicitly represents human de-
cisions on where to develop grazing systems, spans a gradient of identifiable en-
vironmental conditions presented in this section. We show that managed grazing
dominates in the marginal bioclimatic and edaphic regions of drylands. Grazing
occurs in the best bioclimatic areas of temperate forests and woodlands but is
employed on marginal soils found throughout much of the humid tropics.
264
16 Oct 2004 10:42
AR
ASNER ET AL.
AR227-EG29-08.tex
AR227-EG29-08.sgm
LaTeX2e(2002/01/18)
Figure 2 (left) Country-level variables used in a series of Pearson product-moment correlation analyses. (right) Correlations (r-values)
of particular importance to determining linkages between land area, stocking rate, and human population size (all p-values < 0.05). All
unreported correlations were weak and statistically insignificant. Derived by combining References 1, 2, and 8.
P1: GCE
GIS analyses of global grazing by biome reveal that savannas, grasslands, shrub-
lands, and deserts support the largest extent of managed pastoral systems (Figure
3). In combination, these dryland biomes cover more than 67 M km2 of the ∼132
M km2 total global biome area (Table 1). Although these biomes contribute about
51% of the total land area on the planet, they support a disproportionately high
78% of the global grazing area.
Outside of these dryland systems, other biomes support substantial levels of
managed grazing. Roughly 30% and 56%, respectively, of temperate deciduous and
temperate evergreen broadleaf forests and woodlands are now supporting grazing
systems (Table 1); however, some studies indicate that the extent of grazing systems
is decreasing in temperate forests following historically higher levels (4). About
1.7 M km2 (or 10%) of tropical evergreen broadleaf forests have been cleared for
managed grazing, and this area is growing annually in regions such as the Amazon
basin, Congo, and Southeast Asia (5). Many believe that humid tropical ecosystems
represent the only viable way to expand global grazing systems beyond its current
geographic extent (6).
We sought to uncover the environmental conditions under which managed graz-
ing occurs globally. Actual evapotranspiration (AET) was selected as an integrat-
ing metric of bioclimatic stress for vegetation growth, because AET is low in cold
and/or dry regions and high in warm and/or wet areas. We analyzed the spatial
extent of managed grazing relative to the AET for each biome (Figure 4a). There
is a clear bioclimatic footprint throughout global pastoral systems, evident even
in using the coarse analyses afforded by available global GIS data (Table 1). In
savannas, grasslands, and deserts, grazing systems persist in areas where AET is
20% to 25% lower than the average AET of each biome. Managed grazing is pref-
erentially employed in areas that are much drier than the biome mean, as shown in
Table 1, with the ratios of grazed biome:total biome annual precipitation ranging
from 0.69 to 0.82 for savannas, grasslands, and deserts. These results suggest that
the bioclimatically marginal portions of drylands service the global grazing enter-
prise and that other forms of land use (e.g., agriculture and urbanization) occupy
the fraction of dryland biomes with less bioclimatic stress. Indeed, by overlaying
the global extent of croplands (7) on the AET map (8) shown in Figure 4a, we
found that agriculture persists on the fraction of dryland biomes with 47% to 203%
higher AET than the biome average (map not shown).
In contrast to drylands, AET rates of grazing systems in temperate deciduous,
evergreen, and mixed forests are 30% to 75% higher than their biome mean AET
(Table 1). This trend persists in the extreme cold boreal and tundra regions. The
managed grazing footprint is evident as grazing AET:mean biome AET ratios are
from 1.20 to 1.75, most of which is explained by the preferential use of warmer
regions for grazing practices in these cold ecoregions. Temperature conditions in
the portion of boreal evergreen forests used for managed grazing are more than
300% warmer than the biome mean annual temperature. However, these are the
extreme cases; boreal and tundra biomes support less than 1% of the global grazing
enterprise (Table 1).
TABLE 1 Biome statistics of land area, managed grazing, and climatology 266
Total Percent Area Percent Mean Mean Grazing: Grazing: Grazing:
16 Oct 2004 10:42
area of global grazed biome grazing biome biome biome biome

Biomea (M km2) land area (M km2) grazed area AET AET AET MAPb MAT
AR
Savanna 19.31 15 9.48 49.1 595 781 0.76 0.69 0.90

Grassland/steppe 14.22 11 7.68 54.0 321 401 0.80 0.75 0.59
Desert 15.45 12 1.97 12.8 71 88 0.81 0.82 0.95
ASNER ET AL.
Dense shrubland 6.01 5 2.73 45.4 314 339 0.93 0.90 0.96
Tropical evergreen 17.43 13 1.72 9.9 1114 1141 0.98 0.97 0.87
forest/woodland
Temperate broadleaf 1.26 1 0.71 56.0 821 818 1.00 1.00 1.00
AR227-EG29-08.tex
evergreen forest/woodlands
Tropical deciduous 5.96 5 1.20 20.2 935 859 1.09 1.04 1.04
forest/woodland
Boreal evergreen 6.36 5 0.08 1.2 424 354 1.20 0.78 3.03
forest/woodland
Open shrubland 12.09 9 3.98 32.9 297 243 1.22 1.22 1.26
AR227-EG29-08.sgm
Boreal deciduous 2.18 2 0.02 1.1 435 352 1.24 1.45 0.45
forest/woodland
Temperate deciduous 5.10 4 1.49 29.1 793 611 1.30 1.62 1.32
forest/woodland
Temperate needleleaf 3.62 3 0.76 20.9 689 463 1.49 1.59 2.23
evergreen forest/woodland
Evergreen/deciduous 15.68 12 1.26 8.0 642 369 1.74 1.50 1.67
forest/woodland
LaTeX2e(2002/01/18)
Tundra 7.32 6 0.17 2.3 431 247 1.75 1.74 0.01

a
Biomes are ordered by the ratio of actual evapotranspiration (AET) in grazed portions of each biome compared to the biome mean AET. Source data: Ramankutty & Foley
(160), Hearn et al. (8), and Goldewijk et al. (1).
b
P1: GCE
Other abbreviations are MAP, mean annual precipitation, and MAT, mean annual temperature.
Mean annual AET is a metric of the average bioclimatic conditions of an ecosys-

tem, but the manageability of grazing systems (and other forms of land use) is also
largely determined by climate variability, which is not captured in a mean AET
estimate. Regions with large interannual precipitation and temperature variation
undergo climatic boom-bust cycles that strongly affect vegetation production, graz-
ing capacity, and human living conditions. It is extremely difficult to assess the
global footprint of grazing systems with respect to climate variability because
long-term, spatially explicit climate records are not readily available. We used
a 17-year satellite record of the normalized difference vegetation index (NDVI)
(monthly temporal resolution; 1◦ × 1◦ spatial resolution) to map the global inter-

annual variability of vegetation greenness, which is highly correlated with primary
production (9). We calculated absolute NDVI anomalies from the average annual
climatic cycle, 1982–1999. We then compared this measure of vegetation variabil-
ity to the global distribution of managed grazing (Figure 4b). This analysis shows
that global grazing systems persist in nearly all regions with high NDVI variation
(darker gray). Biomes with the highest NDVI variation are, in descending order of
variance: savannas, temperate deciduous forest, shrublands, grasslands, and boreal
systems. Within these biomes, grazing systems occur in zones that are 5% to 63%
more variable in terms of vegetation cover and condition than the mean biome vari-
ability. Managed grazing is thus practiced in the biomes and within the regions of
these biomes that experience substantial climatological and ecological variation.
Managed grazing systems also have a distinguishable global edaphic footprint.
Soil types are shown by taxonomic order (10) along with global grazing extent in
Figure 4c. We calculated the statistical mode of soil type presence by taxonomic
order for grazed areas of each biome and compared it to the mode of soil presence
for the entire biome (Table 2). In savanna, shrubland, and desert biomes, grazing
systems are predominantly found on marginal soils, such as aridisols and entisols,
relative to the most common soils, alfisols, found globally in these biomes. In the
colder boreal biomes, grazing takes place preferentially on alfisols and spodosols
but not on frozen gelisols, which are the most common soil order found in these
regions.
There also exists a clear edaphic footprint of grazing in humid tropical regions.
Ultisols dominate grazing systems found in the Amazon basin, Congo, and South-
east Asia (Figure 4c), yet the most common soils found in these regions are oxisols
(Table 2). Oxisols are widely recognized as nutrient poor and thus marginal for
managed grazing systems (11). Ultisols, are more manageable in terms of fertility
but are also often considered biogeochemically marginal. This global footprint
of grazing systems is probably not a coincidence because ranch managers often
select the best available soils (ultisols) (6). It is nevertheless surprising to observe
a global footprint of managed grazing on soils in the humid tropics, as this area
represents the summed effect of millions of ranch managers operating at small
geographic scales throughout the world. Tropical deforestation is largely driven
by an increasing need for grazing land, with grazing systems expanding in the
humid tropics at a rate of >15,000 km2 year−1 (12). The need for additional
268
16 Oct 2004 10:42
AR
ASNER ET AL.
AR227-EG29-08.tex
AR227-EG29-08.sgm
Figure 4b The global distribution of managed grazing systems and the interannual variability of vegetation production, as indicated by the
satellite metric normalized difference vegetation index (NDVI). Mean NDVI deviation is the interannual variability of vegetation greenness,
after accounting for mean monthly greenness. This record represents the period from 1982 to 1999. Derived from combined References 1
LaTeX2e(2002/01/18)
and 161.
P1: GCE
TABLE 2 Global distribution of managed grazing systems by biome and taxonomic soil order
Mode of soil
order of Mode of
a
Biome grazing areas soil order

Tropical evergreen forest/woodland Ultisols Oxisols
Grazing on the more fertile soil
Tropical deciduous forest/woodland Ultisols Oxisols
Temperate broadleaf evergreen Ultisols Histosols
forest/woodlands
Temperate needleleaf evergreen Alfisols Alfisols
forest/woodland
Temperate deciduous forest/woodland Inceptisols Inceptisols

Boreal evergreen forest/woodland Alfisols Gelisols 

Boreal deciduous forest/woodland Spodosols Gelisols Grazing on the unfrozen soils

Evergreen/deciduous mixed Inceptisols Gelisols 

forest/woodland
Grassland/steppe Mollisols Mollisols

Savanna Entisols Alfisols 


Dense shrubland Aridisols Alfisols 
Grazing on the less fertile soils
Open shrubland Aridisols Alfisols 



Desert Aridisols Entisols
a
Source data: Klein Goldewijk et al. (1), Ramankutty & Foley (160), USDA (10).
grazing land results in part from human population growth but also from pasture
degradation caused by the dominance of low fertility soils (see the deforestation
section).
These GIS analyses illuminate the climatic and edaphic factors limiting the
expansion of grazing systems. Considering the current extent of grazing and a
growing population, expansion of grazing systems in arid and semiarid regions
will require the conversion of cropland systems, a land-use change that is unlikely
to occur given the pressure for grain production worldwide (13). Intensification of
animal production and grazing systems is likely to continue, requiring expensive
management or causing greater degradation of already marginal lands. Any further
extensification of global grazing systems will likely occur through the conversion
of forests to pastures, as is well under way in the humid tropics.
Regional Syndromes in the Global Grazing Footprint

At the global scale, endogenous environmental conditions set the stage for deter-
mining both the ecosystem responses to grazing and the limitations imposed by
climate and soils on the expansion and intensification of grazing practices world-
wide. The bioclimatically marginal nature of grazing lands in arid and semiarid
270 ASNER ET AL.

Figure 5 Three regional syndromes resulting from managed grazing practices across
global-scale gradients of bioclimatic and edaphic conditions.
regions plays a key role in two regional syndromes widely reported throughout the
literature—desertification and woody encroachment. The marginal biogeochemi-
cal nature of humid tropical soils accelerates the process of deforestation, a third
regional syndrome (Figure 5).
One problem in describing these syndromes lies in the definitions of deserti-
fication, woody encroachment, and deforestation. The most difficult to define is
desertification. In the past, a reduction in net primary productivity (NPP; vege-
tation growth) has been used as an indicator of desertification (14). Others have
focused more on the composition and structural configuration of vegetation types
(15), whereas the common observation of increased bare soil (exposed and eroded
surfaces) has been the most definitive trait in yet other studies (16). Ash et al. (17)
argue that desertification can best be analyzed in terms of secondary production
losses (e.g., cattle, sheep, and human). In contrast, woody encroachment has been
defined as the increased geographic extent of woody vegetation in ecosystems.
Expansion of woody vegetation may or may not be a component of desertification;
and when it is, the structural features of the woody cover changes are different from
that of woody encroachment. Most notably, in woody encroachment, herbaceous
cover in the intercanopy zones is typically left intact, whereas in desertification,
these zones become bare soil surfaces with decreased soil resources (e.g., organic
matter). Deforestation has been described as the conversion of forest landscapes
to grazing systems (pastures), but this use is complicated by variation in the def-
inition of forest. Some focus on primary or mature forest; others include areas of
secondary or regrowing forest.
To present our synthetic perspective on the common syndromes inherent to

grazing systems worldwide, we simply define all three processes, desertification,
woody encroachment, and deforestation, in terms of changing ecosystem structure
(Figure 5). Desertification is the replacement of herbaceous cover by shrub cover
and bare soil. Woody encroachment is the addition of woody canopies without
major losses of herbaceous cover, although herbaceous production may decrease.
Deforestation is operationally defined as the replacement of forest cover with
herbaceous pasture systems.
Independent of the precise definition of each regional syndrome, the bioclimatic
and edaphic conditions under which managed grazing occurs have, to some degree,
contributed to the development of these three syndromes. It is widely understood
that desertification has occurred in arid regions of the world (e.g., southwest United
States, Australia, South Africa, and Argentina) as a result of large-scale grazing and
pronounced climatic variability (Figure 5) (3). Woody encroachment, as defined

above, has occurred in semiarid to mesic environments as a result of large-scale
grazing, fire suppression, and climatic variability (18). Deforestation continues to
expand in the humid tropics (and elsewhere) in part because of grazing development
on infertile soils that often cannot sustain large-scale managed pastoral operations
(6). These three syndromes are regional in nature, but they are present throughout
grazing systems on a global scale.
The remainder of this review provides a synthetic perspective on the ways that
managed grazing has altered ecosystems across a bioclimatic gradient from arid
to mesic to humid conditions. Synthesis of vegetation structural and the biogeo-
chemical, hydrological, and atmospheric effects of grazing systems are presented
using the syndromes as the organizing framework. In doing so, we demonstrate how
managed grazing systems contribute significantly to global environmental change.
DESERTIFICATION
Ecosystem Structure
The myriad perspectives and intended audiences of studies on grazing systems lead
to variation in vegetation classifications and, thus, in the observations, analyses,
and conclusions in the literature. Most published studies use one or very few
classifications to describe vegetation-grazing interactions. Common classifications
in grazing studies are woody versus herbaceous (19) or perennial versus annual
(20). Relatively few studies break the vegetation down into classes of C3 versus C4
physiology (21), evergreen versus deciduous life forms (22), or by nitrogen fixing
abilities (23). Few studies have focused at the species level (24), and those that
do often involve the introduction or spread of invasive plants (25, 26). Drawing
from the scientific literature, managed grazing appears to play a central role in
altering the biophysical structure of grazed ecosystems globally. We use ecosystem
structure to discuss the spatial extent and configuration of major vegetation life-
forms (trees, shrubs, and herbaceous cover). This is useful because the majority
272 ASNER ET AL.
of studies describe changes in ecosystem structure that appear to be caused or

accelerated by grazing practices.
The literature highlights a consistent set of ecosystem structural changes in-
volved in desertification. These changes can be described in three major features
and one overarching pattern. The features include (a) increased bare soil surface
area, (b) decreased herbaceous cover, and (c) increased cover of woody shrubs and
shrub clusters. The overarching pattern is one of increased spatial heterogeneity of
vegetation cover and a concomitant increase in the spatial variance of belowground
resources, such as organic matter, nutrients (see the Biogeochemistry section), and
soil moisture (see the Hydrology section).

Many grazing systems experiencing desertification in the southwestern United
States, Australia, and Africa are now dominated by one or a few woody shrub
species, with little herbaceous canopy remaining on the landscape (27–29). Okin
et al. (16) suggested wind erosion removes soil nutrients and carbon from shrub
interspaces. Once established, a combination of biogeochemical and hydrological
feedbacks sustains these shrub systems in a new stable state, very different from
the prior grassland (Figure 6) (15).
Desertification can also happen without a major increase in woody plant cover
but rather as an increase in bare soil causing fragmented herbaceous cover (30).
Van de Koppel et al. (31) suggested the following progression of events leading
to a fragmented landscape. Herbaceous cover decreases in areas preferential to
Figure 6 Processes mediating desertification in arid grazing systems.

grazers, leaving compacted bare soils that allow rainfall to run off into remaining
vegetation patches. This increases the productivity of remaining patches, which
stimulates increased grazing and, thus, loss of these patches, eventually leading
to total ecosystem collapse. There is currently little empirical evidence to test this
model, but another modeling study concurs that patch dynamics are mediated by
grazing, climate variability, and surface hydrological transport (32).
A phenomenon related to managed grazing, land degradation, and desertifica-
tion is the human-mediated dispersal of African grasses worldwide. Introduced
African grasses have made their ecological mark in dryland (and tropical) sys-
tems in North America, Central and South America, Australia, and Oceania (33).
These grasses compete effectively with native grass species and can alter nutrient
cycling and other ecosystem processes (34–36). African grasses are typically fire
tolerant and quite flammable, increasing fire frequency and promoting their further
geographic expansion (26).
Biogeochemistry
Some changes in ecosystem structure (abundance, cover, and configuration of
life-forms) described above are directly attributable to grazing. These structural
alterations result in a cascade of change in other ecosystem processes, such as
water drainage, wind and water erosion, species invasion, disturbance types and
frequency (e.g., fire), carbon cycling, and the biophysical and biochemical char-
acteristics of soils (21, 31).
Where desertification is occurring, degradation often results in reduced pro-
ductivity or vegetative cover, which brings with it a change in the carbon (C) and
nutrient stocks and cycling of the system (Figure 7) (37–43). The primary ecosys-
tem response to desertification is an increase in the heterogeneity of vegetation
cover, with concomitant increases in the spatial variability of soil C and nutrients
(15, 54). Overall, reduced vegetative cover and total aboveground biomass seem
to result in a small reduction in aboveground C stocks and a slight decline in C fix-
ation, measured as NPP, but there is significant variability by vegetation type with
topographic and edaphic factors (43). Because total nutrient pools are relatively
small, any decline in total nutrient stocks has a significant impact on productivity.
Despite small, sometimes undetectable changes in aboveground biomass and
NPP, both total soil C and nitrogen (N) usually decline (38). The reduced soil
organic N may result from increased N lost in surface runoff, increased trace-gas
flux, and vegetation removal by grazers. Decreased infiltration and increased runoff
elevate losses of both inorganic and organic N in overland flow; however, these
losses are smaller than inputs to the system from deposition in some regions (39,
55). Nitrogen trace-gas losses, particularly as nitrous oxide, are large relative to the
total N pool (see the next section). Most studies on biogeochemical changes caused
by desertification are from Northern Chihuahuan ecosystems in New Mexico,
United States of America. A recent study by Asner et al. (37) in Argentina also
found that desertification resulted in little change in woody cover, but there was
274
16 Oct 2004 10:42
AR
ASNER ET AL.
AR227-EG29-08.tex
AR227-EG29-08.sgm
LaTeX2e(2002/01/18)
P1: GCE
a 25% to 80% decline in soil organic C and N storage in areas with long-term
grazing.
Soil compaction reduces infiltration and increases runoff, resulting in faster
and greater flow through waterways and greater channel and gully erosion (see
the Hydrology section). As a result, soil is lost, and sediment loads increase in
waterways (56). In systems where soil erosion is substantial, resulting losses of
soil C, N, and P may also be important but have not been well quantified. Wind
erosion could also be important and enhanced by increased bare soil in the case
of desertification (16). Phosphorus in arid ecosystems is often bound to calcium
carbonates and retained in the mineral soils; thus, erosion tends to mobilize C
and N more so than it does P (57). Grazing also breaks up and reduces coverage
of cryptobiotic soil crusts found in many arid regions. Although disturbing these
crusts can increase infiltration of water, it also reduces inputs of C and N fixed by
biological activity of the crusts (58).
Trace-Gas Emissions
Increases in greenhouse trace gases (CO2, CH4, N2O, and O3) since preindustrial
times have led to a warming of the Earth’s surface and other climate changes
(59). Though anthropogenic emissions of trace gases from fossil-fuel burning and
fertilizers account for the majority of trace-gas emissions, soils account for more
than 30% of biogenic trace-gas emissions (60). Soil trace-gas production and
consumption vary spatially and temporally, and they are governed by factors such
as soil nutrient stocks and cycling rates, soil temperature and moisture content,
and vegetation cover (61), all of which are changed by grazing (previous sections).
Nitric and nitrous oxide gases (NO and N2O) are produced in the soil during
the processes of nitrification and denitrification, and these fluxes are mediated
by vegetation litter inputs. The partitioning of NO and N2O fluxes at the soil-air
interface is dependent upon the soil water content, with a shift from NO production
to N2O production as the soil water increases (62). Nitric oxide is a key component
in regional-scale ozone regulation.
Woody vegetation cover changes associated with desertification have measur-
able impacts on soil NO emissions. Hartley & Schlesinger (40) found higher NO
emissions from soils under woody canopies than in intercanopy zones. These find-
ings support the concept of enhanced nutrient stocks and cycling under woody
vegetation canopies, thereby resulting in enhanced N gas emissions at the plot
scale. Although there were little to no N2O emissions measured in the field from
←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
Figure 7 Effects of three syndromes of managed grazing on biogeochemical prop-
erties of arid/semiarid, mesic, and humid tropical ecosystems. Carbon (C), nitrogen
(N), and phosphorus (P) stocks and fluxes are taken from literature sources cited to the
left. All pools are in kg/hectare (ha) except those noted as a percentage, and all fluxes
are in kg/ha/year except for the tropical systems where they are kg/ha/burn. Values for
tropical ecosystems are averages across multiple studies.
276 ASNER ET AL.
bare soils due to the aridity, laboratory studies indicated N2O emissions from soils
collected under shrub canopies can potentially be two times greater than those
collected in a nearby grassland (63).
Methane (CH4) is generated in soils during anaerobic respiration, and it has a
global warming potential 24.5 times that of CO2. In some cases, soil compaction
due to grazing limits soil aeration and stimulates CH4 production (64). Grazing
affects the atmosphere through the direct emission of CH4 gas from ruminants
and via increases in ammonia (NH3) production from livestock excreta. Globally,
approximately 54 terragrams (Tg) N-NH3 are emitted each year, with the largest
fraction (∼40%) from animal waste (65). Excreta deposited on grassland by graz-
ing animals stimulated N2O production, contributing up to 22% of the total N2O
emission from a U.K. grassland (66). The current estimate of CH4 from ruminant
animals and animal waste is 100 Tg CH4, nearly a fifth of the total global emis-
sions (59). CH4 emissions from livestock have increased about fivefold over the
last century in close step with the increasing rate of cattle production (67).
Hydrology
The balance between plant-available water and evapotranspiration (ET) regulates
soil moisture and ultimately determines many characteristics of ecosystem struc-
ture and functioning on grazed lands (68). In grazing systems, ET (calculated as
the sum of evaporation from bare soil and plant surfaces and from transpiration
through plant stomata) is often the largest loss of water from the system (Figure 8)
(56, 69, 70). Controls over this flux are the relative fractions of soil and vegetation
cover, leaf area index (LAI), and plant-available soil moisture. Grazing influences
each of these variables.
Vegetation cover and LAI decline as grazers remove plant matter. Reduced
plant surface area results in lower transpiration and retention of soil moisture
throughout the root zone (Figure 8) (71). Grazing thus leads to increased soil
Figure 8 Components of the hydrologic cycle and reported directions of change with
managed grazing. Processes are divided into aboveground(AG) and belowground(BG)
components. Other abbreviations are PR, rainfall; PS, spring snow melt; F, fog or
cloud condensation on aboveground plant matter; I, canopy interception; R, runoff;
E, evaporation from soil surfaces; T, transpiration from the canopy; S, the change in
soil moisture; and D, discharge through subsurface flow vertically and horizontally
away from plant roots. PS and F are not applicable in all environments.
moisture relative to nongrazed pastures (72–74). Grazing also increases bare soil
surface area, resulting in greater radiative heating of soil surfaces and increased
evaporation (71). Therefore, net changes in soil moisture can be assessed with
regard to the balance between LAI and bare soil surface changes resulting from
grazing. Reductions in LAI alone will increase soil moisture; increases in bare soil
area will decrease soil moisture; and when both occur, the change in soil moisture
is uncertain (75).
Grazing compacts soil and exposes soil surfaces, both of which lead to lower
infiltration, increased runoff, and higher erosion rates (Figure 8) (56). Trimble &
Mendel (76) found that infiltration decreased from approximately 50 mm h−1 on

lightly grazed to 25 mm h−1 on heavily grazed land, but they also highlighted
the large variance in field measurements. Infiltration is sensitive to many factors,
including soil conditions at the time of grazing and the degree of bare soil ex-
posed following grazing. The link between higher runoff and erosion rates follows
logically because increased surface flow can carry larger sediment loads. Graz-
ing reduces vegetation cover, leading to bare and unstable soil surfaces; there-
fore, the effects of grazing on erosion can be notably larger than the impact of
climatic changes, such as increased precipitation (77). Conversely, the absence
of grazing results in litter buildup, which has been found to reduce runoff and
erosion (78). The balance between infiltration and runoff depends on the hy-
drologic conductivity and spatial heterogeneity of soil and vegetation surfaces
(79).
Desertification involves specific hydrologic changes that include increased spa-
tial heterogeneity of soil moisture and less effective transfer of precipitation to soil
moisture (15). Net primary production per unit of precipitation may decrease (80),
creating landscapes, with a low rain-use efficiency, that resemble deserts. There
is evidence that grazing, combined with feedback mechanisms in the climate sys-
tem, eventually makes these symptoms permanent (31). The process may begin in
areas with vegetation cover removed and soils compacted, resulting in declining
infiltration rates and moisture in surface soils. Deeper soil layers and soils where
vegetation cover remains high may continue to receive recharge from large storm
events (81). The landscape thus becomes more heterogeneous, with patches of veg-
etation helping to maintain higher infiltration, soil stability, and nutrient retention.
Meanwhile, bare soil interspaces become increasingly depauperate in soil mois-
ture and nutrient resources. With continued grazing and high climate variability,
vegetated areas become increasingly rare, shifting systems even further toward a
desert-like state.
Climate Interactions
Early studies of desertification suggested that changes in surface albedo caused
by increased bare soil cover could have regional and even global climate effects.
Charney et al. (82) predicted that increases in surface albedo due to Sahelian
desertification would increase radiative heat losses from the Sahara, thus reduc-
ing rainfall. Additional studies incorporating albedo, transpiration, and roughness
278 ASNER ET AL.
found that a positive feedback reduced rainfall in the Sahel as well as on the Indian
subcontinent (83).
In Sonoran drylands, surface temperatures were generally 2◦ –4◦ higher on the
brighter, more heavily grazed Mexican side of the border than on the U.S. side
(84). In this case, temperature differences were shown to impact soil moisture
and cloudiness, but no changes in precipitation were apparent. A more recent
study, using Landsat Thematic Mapper data, found only small Arizona/Sonora
trans-border differences in albedo and radiant temperature along 25 one-kilometer
transects (85).
WOODY ENCROACHMENT
Ecosystem Structure
There are hundreds of documented cases of increased woody plant cover in

semiarid, subtropical rangelands of the world [(86); http://cnrit.tamu.edu/]. In
North and South America, Africa, Australia, and elsewhere, woody vegetation
cover has increased significantly in grazing systems during the past few decades.
Cited causes of woody encroachment include overgrazing of herbaceous cover that
reduces competition for woody seedlings, fire suppression that enhances woody
plant survival, atmospheric CO2 enrichment that favors C3 (woody) plant growth,
and nitrogen pollution that favors woody encroachment (Figure 9) (18, 87).
Figure 9 Processes mediating woody vegetation encroachment in semiarid and mesic

grazing systems.
It is noted in most encroachment studies that the woody plants were present
somewhere on the landscape prior to the installment of managed grazing. For
example, in a southern Texas rangeland containing a diverse array of trees, shrubs,
and subshrubs, heavy grazing caused increases in the cover of the nitrogen-fixing
tree Prosopis glandulosa var. glandulosa (mesquite). Long-term records and aerial
photographs indicate that mesquite encroachment then facilitated the establishment
of other woody plants in its understory, which subsequently outcompeted mesquite
for light and other resources (88). Mesquite remnants are commonly found among
well-developed patches of woody vegetation known not to have existed a century
ago (89). The same species of mesquite has increased dramatically in cover in
a northern Texas rangeland during the past century (44), but there are very few
other woody species established in this region. Most other species are confined to
riparian zones; thus few woody plants can be found in association with the mesquite
cover (90). Precipitation conditions are similar between the northern (650 mm)
and southern (680 mm) Texas sites, but temperatures are substantially lower in the
north, with values below freezing in many months (91). Low temperatures in the
north likely preclude the presence of many warm-climate woody plants found in
the south (Acacia, Diospyros spp.), and thus the ecological dynamics of woody
encroachment are very different between sites.
There are some basic trends in vegetation-grazing interactions associated with
woody encroachment in global drylands. Five vegetation properties are consis-
tently highlighted in the literature as changing with respect to grazing and/or
the release from grazing: (a) woody vegetation cover, (b) herbaceous vegetation
cover, (c) surface litter cover, (d) dominance by perennial herbaceous plants, and
(e) dominance by annual herbaceous plants (Figure 10).
At light grazing intensity, most studies indicate slight increases in woody cover
if the woody plants are present in the area (Figure 10) (92). Somewhat independent
of woody vegetation dynamics, most reports show a decrease in herbaceous veg-
etation biomass and/or cover and in surface litter cover/biomass (24, 93). In light
grazing scenarios, some studies mention decreases in perennial grasses, although a
clear trend among annuals is not evident (94). Changes are much more pronounced
in cases of long-term, heavy grazing (Figure 10).
Many studies indicate dramatic increases in woody cover or biomass if the
woody plants are already present or introduced to the region (86). Both herbaceous
and surface litter (cover and biomass) are found to decrease under conditions of
heavy grazing (37, 78). Shifts from perennial to annual grasses are more obvious in
heavy grazing regimes (95, 96). Following heavy grazing, a release, or substantial
rest period, woody cover often remains elevated (Figure 10); that is, there are few
if any studies showing decreases in woody cover following release from heavy
grazing, but herbaceous and surface litter cover and biomass typically do increase
(97). A few studies indicate that annual grasses may initially increase following
release from grazing, but often the grasses are replaced by perennials over periods
of years or a few decades (98). A net outcome of heavy grazing, even following a
release from such practices, can be an increase in woody cover regionally.
280 ASNER ET AL.

Figure 10 Commonly reported responses of five-dryland ecosystem structural properties

to light and heavy grazing and to release from grazing. Bars show relative, directional
responses as reported in the literature.
Biogeochemistry
Woody encroachment reduces the quality of land for animal production (39), yet
in some cases, it enriches total ecosystem C and N stocks (Figure 7). As previously
discussed, the shift from herbaceous to woody vegetation is different from deser-
tification in that the woody cover increases without a major loss of herbaceous
cover. With the shift to woody vegetation comes a large increase in aboveground
NPP and C storage. Increases in aboveground NPP of up to 1400 kg C ha−1 year−1
have been observed when the dominant woody species is a nitrogen fixer (48, 90).
Increases in the aboveground C pool can range from 300 to 44,000 kg C ha−1
in less than 100 years of woody encroachment (44). When the dominant woody
species is a nitrogen fixer, nitrogen accumulation can be 9–40 kg N ha−1 year−1
greater in the woody areas than the grasslands (48), with aboveground nitrogen
increasing 39–468 kg N ha−1 following encroachment (46).
Despite an increase in aboveground C and N with encroachment, the trends
in soil organic C and N are highly variable. Measuring soil carbon to 3 m depth,
Jackson et al. (45) found that woody encroachment increased soil C and N in
drier grassland regions but reduced it in regions with mean annual precipitation
greater than ∼500 mm. They found the decline in soil C in wetter ecosystems
was sufficient to offset aboveground gains from woody encroachment, resulting in
no net ecosystem C gain. In contrast, Boutton et al. (99) studied grasslands with
annual precipitation of 700 mm and found a 27% to 103% increase in soil C in the
upper 10 cm. Likewise, Geesing et al. (48) measured soil organic C increases of
40% to 80% across a range of sites with mean annual precipitation of ∼700 mm.
As with soil C, there is some uncertainty in how total soil N changes with
woody encroachment. Jackson et al. (45) found trends similar to those observed
for C, an increase in N stocks in drier regions and a decrease in wetter regions.
In contrast, Hibbard et al. (47) found a clear increase in soil N in the top 10 cm
of a wetter site, and Martin & Asner (49) measured an increase in soil N pools
following 30–70 years of woody encroachment. Clearly, our understanding of how
woody encroachment changes soil resources, and thus the long-term productivity
and sustainability of semiarid regions, remains highly fragmented.
Trace-Gas Emissions
Woody encroachment has a measurable effect on soil nitrogen oxide emissions. In

northern Texas rangelands, encroaching Prosopis glandulosa, a N-fixing species,
caused C and N storage to increase in surface soils, which resulted in enhanced
soil nitric oxide (NO) fluxes during nitrification (49, 50). Aboveground woody
biomass was the best spatial predictor of NO emissions, with values increasing
20-fold (0.04–0.78 mg NO-N m−2 day−1) across a 70-fold biomass gradient (5–
350 g m−2). Emissions also covaried with soil pH and clay content. Temporally, NO
emissions and nitrification were positively correlated with temperature. Precipi-
tation events elevated NO emissions fourfold over 24-hour periods and produced
small amounts of N2O. Overall, mesquite encroachment in these grasslands in-
creased NO emissions in a spatially explicit manner determined by the woody
biomass and soil type, which was then temporally mediated by temperature and
secondarily by precipitation (49, 50).
At a regional scale, desertification and woody encroachment appear to have very
different effects on the N status of dryland ecosystems. Desertification promotes
nutrient accretion in soils under woody plant canopies, but the surrounding bare
soil areas have very low-nitrogen contents. When the fractional covers of woody
clusters with higher NO emissions and bare soils with low-NO emissions are taken
into account, there is ∼0.4 kg N ha−1 year−1 (53%) decrease in NO emissions re-
gionally, in comparison to the preexisting grassland (40). In contrast, when woody
encroachment involves an N-fixing species in a grassland, and when woody expan-
sion does not decrease the herbaceous cover significantly, NO gas emissions in-
crease beyond that of the original grasslands by ∼1.0 kg N ha−1 year−1 or 29% (49).
Hydrology
Woody and herbaceous life-forms utilize soil moisture from different depths in
the soil profile (100, 101). Woody plants take advantage of deeper soil moisture,
and herbaceous plants access moisture only in the upper soil layers. It has often
been theorized that variation in soil moisture, both vertically and horizontally,
determines the relative fraction of woody and herbaceous cover (102). Grazing
282 ASNER ET AL.
affects soil moisture through compaction and reduced infiltration and through the
exposure of bare soil surfaces. Increased evaporative losses from bare soils increase
the disparity between shallow and deep soil moisture, and such losses support the
notion that surface soil moisture is more sensitive to grazing intensity than deep
soil moisture (103). These changes in soil moisture may drive changes in the
relative balance between woody and herbaceous cover, favoring the deep-rooted
shrub species.
Woody encroachment may be controlled by changes in the annual timing of
precipitation (32) or the relative proportion of rainfall occurring in large precipi-
tation events (100). Particularly in the presence of grazing, high-rainfall events or

several months of elevated precipitation are effective in recharging deep soil lay-
ers, thereby creating soil moisture conditions that favor woody species. Modeling
studies confirm the importance of temporal precipitation patterns in determining
the relative abundance of herbaceous and woody plants, but grazing activity must
be included for full transitions between plant life-forms in these models (32).
At the scale of individual canopies, shading, litter accumulation, and canopy inter-
ception of precipitation causes soils beneath woody vegetation to receive less solar
irradiance and to have lower temperatures and water contents compared to inter-
canopy areas (104, 105). However, recent work suggests that the albedo changes
resulting from shrubland encroachment have been slight. For example, Grover &
Musik (106) showed that mean daily albedo for a creosote bush shrubland and a
grassland were not significantly different (0.2 and 0.27, respectively), and these did
not differ from other creosote bush, mesquite, or grassland communities. They con-
cluded that increasing spatial heterogeneity of woody cover does not significantly
impact surface albedo and climate. In contrast, Hoffman & Jackson (107) attributed
a decline in precipitation of approximately 10% to a reduction in roughness length
and to an increase in albedo. They also found that deeper rooting had a small
positive effect on latent heat flux, with a corresponding reduction in sensible heat.
In comparison to canopy-scale processes, the impacts of woody encroachment
on regional climate variables are virtually unknown. It is thought that the effects of
woody encroachment on surface energy fluxes, critical to the formation of clouds
and precipitation, are small relative to evaporative fluxes from other forms of land
use (e.g., irrigated croplands). However, the spacing of vegetation in semiarid
regions can affect the development of local wind circulations that contribute to
cumulus cloud formation and precipitation (108).
DEFORESTATION
Ecosystem Structure
At first glance, the ecosystem structural changes caused by forest-to-pasture con-
version might seem obvious. Biologically diverse, large-stature forest is cut (and
often burned) in geometric patterns easily discernable from satellite imagery. This
is true, and in humid tropical regions, the “installed” herbaceous species are pre-
dominantly African grasses (33, 109). However, tropical pasture development is
carried out across an enormous range of environmental, social, and economic con-
ditions, leaving the installed pasture in an equally broad number of biophysical
states. In the humid tropics, the biophysical structure of pastures ranges from highly
managed monospecific grasslands to savanna-like systems containing varying den-
sities of palm and secondary forest species. In a study of 145 pastures in the central
Amazon, Asner et al. (110) found that shrub and secondary forest vegetation cover
ranged from 0% to 45%, and palms covered 20% to 60% of pasture areas. Al-
though variation in pasture cover causes concomitant variation in biogeochemical,

atmospheric, and hydrological processes (see below), vegetation structure is rarely
quantified in published studies. Therefore, our synthesis implicitly incorporates the
complicating effects of structural variability in pastures, but it does not determine
the effects of this variation.
Biogeochemistry
In forests and woodlands, the largest impacts of grazing result from the conversion
of the system to herbaceous cover, and deforestation for cattle pasture occurs
across a wide range of climatic conditions. We discuss both dry and wet tropical
forest conversion, where future conversion is most likely to take place on a large
scale. With tropical dry forest conversion followed by repeated burning, up to 90%
of aboveground C and N stocks and nearly 50% of the aboveground phosphorus
(P) can be lost (111). C, N, and P are volatilized, whereas calcium (Ca) and the
remaining C, N, and P are deposited as ash, much of which is lost in wind erosion
immediately after the fire. Water erosion and soil loss can be significant for a few
months after fire (112), and concentrations of dissolved mobile nutrients [e.g., NO3
and potassium (K)] can be substantially elevated in overland flow (113). For a short
period after burning, large increases in available soil N and nitrification rates are
observed because NH4 is mineralized during fire (114). Fire has little direct effect
on total soil nutrients (C, N, and P); losses of these nutrients are mainly from the
aboveground pools, and the size of these losses is small relative to the size of total
soil nutrient pools (111, 115). However, it is estimated that it would take a century
or more of recovery for a dry forest ecosystem to accumulate the nutrients lost
during slash and burn (111).
With the global expansion of grazing systems, further conversion is most likely
in humid forests. In contrast to dry forests, humid tropical forests are often found
on highly weathered soils (ultisols and oxisols) rich in available N and poor in P
and base cations [e.g., Ca and magnesium (Mg)], which are weathered and leached
from the original rock (116, 117). Losses of P and Ca during deforestation are thus
important to future pasture productivity (Figure 7). Many studies show that the
initial burning of slashed primary forest results in combustion of ∼48% of biomass,
or 88 Mg C, 1181 kg N, and 107 kg sulfur (S) per hectare (52). The percentage
of aboveground nutrients lost through combustion and transport of particulates
average roughly 90% of N, 45% of P, and over 30% of Ca, Mg, and K (118). For
284 ASNER ET AL.
about a decade after conversion, pastures release significantly more N2O (nitrous
oxide) trace gas than the forest but then decline to background forest levels (see
the Trace-Gas section). Following deforestation, soil pH and exchangeable cations
remain elevated in pastures, but a recent synthesis showed that neither soil C nor
N changed in a consistent manner (11, 119). However, a detailed reaccounting of
changes in soil C that considered the effects of soil compaction and management
practices suggests an average loss of 12 tons C ha−1 in tropical lands maintained
as pasture (53).
Repeated burning is often used as a management tool to remove woody regrowth
and weeds and to renew nutrient availability in pastures (120). In the Brazilian
Amazon, repeated burning of cattle pasture consumes up to 46% of aboveground
biomass (slash, grass, and litter), with ∼14 Mg C, 199 kg N, and 16 kg S per
hectare lost to the atmosphere (52). Losses of P, Ca, and K in one experimental
burn were nearly 33 kg ha−1 (120). With repeated fires over a six-year period, over
1900 kg N ha−1 is lost, which is equivalent to ∼90% of the aboveground pool in a

mature tropical forest. As a result, repeated burning can lead to N limitation even
in previously nitrogen-rich tropical systems (121).
Results from the literature are inconsistent in describing how P changes with
deforestation and repeating burning of pastures (11, 51, 118). Available P in ash
after the initial fire may remain in the system, rapidly sorbing to soil minerals or
consumed by vegetation and soil microbes. In many cases, available P seems to
accumulate in soil organic material, which can remain enriched in pastures for
years after conversion (11, 122–124). Others do not find the missing available P
in the soil (51).
With long-term grazing (e.g., decades), total soil P usually declines; Ca, K, and
Mg often decline as well; and N may also decline if there is frequent burning (44,
118, 125). The mechanisms for P loss are uncertain but include combustion, erosion
of ash, and leaching to deep soils. Another possibility is that P is transferred into
a bound form that is not easily detected with methods currently used. In any case,
available phosphorus declines and grass productivity in pastures is often P limited.
Phosphorus fertilization (∼50 kg ha−1 every 5–10 years) is thus becoming more
common in some regions, such as the eastern Brazilian Amazon (121, 126). Despite
these recent and more localized trends, many studies show that tropical pastures
accumulate and cycle fewer nutrients than forest, redistribute cations from trees to
soils, lose most of the C and N that was stored in aboveground forest biomass, and
maintain reduced soil nutrient availability (Figure 7). These reported trends are
induced and mediated by poor edaphic conditions inherent to many humid tropical
forest regions.
Trace-Gas and Aerosol Emissions

Tropical forest soils are the largest biogenic source of N2O, accounting for 25%
to 50% of the global source (59). NO emissions from tropical forests are also sig-
nificant, accounting for up to 20% of global emissions (127). The disturbance and
initial volatilization of nutrients during burning associated with forest-to-pasture
conversion increases trace-gas emissions for months to years, but emissions often
return to or drop below initial forest or savanna levels as pastures age. Studies
in Costa Rica demonstrated that soil N2O emissions and CH4 production were
higher for 15 years following conversion, but returned to background levels after
18 years (128). In a synthesis, Davidson et al. (129) found for a variety of tropi-
cal sites in Costa Rica, Puerto Rico, and Brazil that old tropical pastures produce
consistently lower NO fluxes than old-growth tropical forests. The reasons for
these differences were changes in the environmental factors that control N oxide
emissions, such as soil water, temperature, nutrient status, pH, diffusion, and plant
biomass.
Biomass burning is a common tool for the establishment and management of
pastures for grazing, most recognized in tropical rain forests and savanna regions
but also in grasslands worldwide (130, 131). Biomass burning is a significant
source of globally relevant trace gases (CO2, NOx, CO, and CH4) and aerosols
(130, 132). Climatological effects include the formation of photochemical smog,
hydrocarbons, and NOx that rapidly produce O3. Long-range transport of smoke
plumes may be redistributed locally, transported throughout the lower troposphere,
or entrained in large-scale circulation patterns in the mid and upper troposphere.
The perturbation of these gases to the atmosphere is evident in satellite observations
of high-O3 and -CO levels over large areas of Africa, South America, and the
tropical Atlantic and Indian Oceans (133).
Pyrogenic aerosols from pasture biomass burning dominate the atmospheric
concentration of aerosols over the Amazon basin and Africa (132, 134). Con-
centrations of aerosol particles are highly seasonal, with a clear maximum in the
dry (burning) season, contributing to cooling both through increasing atmospheric
scattering of incoming light and the supply of cloud condensation nuclei (CNN).
High-CCN concentrations from biomass burning stimulate rainfall production and
affect large-scale climate dynamics (135). The cooling effect of smoke alone may
be minimal (−0.3 watts m−2) compared to the heating from anthropogenic green-
house gases [2.45 watts m−2 globally (136)].
Hydrology
The conversion of forest to pasture is the primary driver of grazing-induced hy-
drologic change in mesic to wet climate zones (137). The majority of research
conducted in tropical environments is in regions where deforestation has occurred
recently and is likely to continue (5). Higher surface albedo, lower surface aerody-
namic roughness, reduced LAI, and shallower rooting depths combine to reduce
ET in pastures relative to forests. This model is supported by stable isotope studies,
which reveal that water vapor above forest sites is derived from plant transpiration,
in the eastern Amazon, whereas water vapor above pasture is derived primarily
from surface evaporation (138). Lower ET and reduced infiltration rates within pas-
tures culminate in increased average long-term discharge (139). Where changes
in infiltration rates are modest, reduced ET alone can increase rainy season runoff
(140). Decreases in dry-season flow are also theorized to result from deforestation,
286 ASNER ET AL.
but evidence of this effect is limited (141). Initial increases in stream discharge
decline over time as forests regenerate (137, 142).
Forest-to-pasture conversion causes substantial decreases in land-to-atmosphere
moisture transport. Early global-scale numerical simulations, using drastic levels
of deforestation, predicted a 25% decrease in precipitation associated with a 30%
decrease in evapotranspiration and a 2.5◦ C increase in surface temperature (143).
However, recent mesoscale modeling studies suggest that, in spatially complex
mosaics of forest and pasture lands, moisture fluxes from the land to the atmosphere
can be enhanced, as can reciprocal fluxes of precipitation (144). Recycling of water
by up to 25% to 35% was attributed to interactions between increased albedo,
sensible heat flux, and mixed-layer height in pastures relative to forest (144, 145).
Modeling studies also suggest that changes in surface energy dynamics can affect
the upper atmosphere, perturbing tropical circulation patterns, shifting the position
of the Hadley circulation, and altering planetary waves that propagate moisture to
upper- and midlatitudes (146). Nonetheless, the effects of managed grazing and
deforestation on climate, like that of desertification and woody encroachment, are
poorly understood.
ECOSYSTEM RESPONSES TO MANAGED GRAZING

Response Typologies
Managed grazing has flourished for thousands of years, but the spatial extent and
intensity (e.g., stocking rates) of grazing systems have increased substantially in
the past several decades to centuries. The typologies of ecosystem response to
managed grazing are regional in nature because they vary with bioclimatic and
edaphic conditions. Not all grazing systems or practices lead to the syndromes of
desertification, woody encroachment, and deforestation highlighted in this synthe-
sis. We have a poor understanding of how much land has been affected by these
phenomena. Nonetheless, scientific reports tend to address common environmental
concerns, and these are the phenomena that often emerge as identifiable syndromes
of managed grazing.
In arid regions, ecosystem responses to grazing practices are mediated by ex-
treme climatic conditions combined with nutrient-poor soils. Low and highly vari-
able precipitation causes ecological boom-bust cycles (Figure 4b), and yet managed
grazing occurs on the most variable and climatologically marginal portions of these
biomes (Table 1). When grazing systems are implicated in cases of desertification,
thresholds in ecosystem resistance and resilience to drought are often crossed ow-
ing to the persistence of grazing at the worst of times climatologically (27, 147).
The results are long-term losses of surface herbaceous cover, increases in bare
soil extent, and, at times, increases in woody shrub cover (Table 3). A cascade
of biogeochemical and hydrological feedbacks then takes place, such as nutrient
TABLE 3 Relative effects of land-cover change due to grazing on land surface

properties that mediate biosphere-atmosphere interactions. Larger arrows indicate
the dominant change within the ecosystem syndromes listed
Deforestation Woody proliferation Desertification
Albedo ↑ ↓ ↑
Roughness length ↓ ↑ ↓
Turbulence ↑ ↑ —
↓ ↑ ↓
Vegetated fraction
Evaporation ↓ — ↑
Transpiration ↓ ↑ ↓
losses via runoff and wind erosion, soil compaction, reduced soil water infiltration
and increased patchiness of soil moisture (Table 4). Structural, biogeochemical
and hydrological processes change to the point where an alternative “stable” state
ecosystem then persists (15, 148). Additional feedbacks to the atmosphere result
from changing albedo, surface temperature, and trace-gas emissions (Table 3).
The scientific community is only beginning to understand the potential effects of
desertification at the ecosystem-atmosphere interface. Whether these changes alter
regional climate in ways that may enhance or dampen the effects of desertification
on the global climate system is not known.
In semiarid and mesic biomes, woody encroachment is a widely reported
ecosystem response to managed grazing (Table 4); however, other critically impor-
tant cofactors are climate variability, fire suppression, and the presence of woody
plant seed sources (Figure 9). Fire suppression favors woody seedling recruit-
ment and survival, especially when grazers are actively consuming the herbaceous
layer. In contrast to desertification, which appears to entail a combination of ex-
treme climate variability and heavy grazing, even light-to-moderate grazing in-
tensities can promote woody encroachment in semiarid and mesic environments
(86, 149) (Figure 10). Like that of desertification, the spatial heterogeneity of
aboveground and belowground resources such as vegetation cover and soil or-
ganic matter often increases. However, woody vegetation increases tend not to be
well linked to large-scale losses of herbaceous cover. Biogeochemical responses
include soil compaction, increased carbon storage and nutrient stocks (especially
when nitrogen fixing woody species are implicated), and increased greenhouse
gas emissions. The hydrology of these systems shows a typological response as
well, with increased heterogeneity of soil moisture with depth, increased runoff
and erosion, and decreased soil water infiltration (Table 3). Virtually nothing is
known regarding the effects of woody encroachment on the climate system.
The major effects of managed grazing in humid forest regions are determined
by deforestation and pasture maintenance (Table 4). Forests are replaced by
herbaceous systems, which decrease ecosystem carbon storage and nutrient stocks
288
16 Oct 2004 10:42
TABLE 4 Summary of most commonly reported changes in ecosystem properties for desertification, woody encroachment, and
deforestation
AR
Regional syndrome
Ecosystem properties Desertification Woody encroachment Deforestation

ASNER ET AL.
Vegetation structure • Decreased herbaceous, • Increased woody cover • Increased herbaceous

increased bare soil • Increased spatial heterogeneity • Decreased forest cover
• Increased woody cover
AR227-EG29-08.tex
• Increased spatial heterogeneity

Biogeochemistry • Increased spatial heterogeneity • Increased carbon storage • Decreased carbon storage
of nutrients and carbon • Increased soil nutrient • Decreased soil nutrient stocks
• Increased nutrient loss via stocks, when N-fixing plants over time
runoff and erosion are present and active • Nutrient losses via burning
• Decreased soil nutrient stocks • Soil compaction and aerosol
• Soil compaction
AR227-EG29-08.sgm
Biosphere-atmosphere • Increased albedo • Increased N trace-gas production • Changed surface energy budget
exchange • Increased surface temperature • Increased ammonia and • Short-lived increase in soil
• Increased ammonia and methane production N trace gas emission
methane production • Increased ammonia and
methane production
• Aerosol production
Hydrology • Increased spatial heterogeneity • Increased vertical heterogeneity • Decreased transpiration
LaTeX2e(2002/01/18)
of soil moisture of soil moisture • Decreased infiltration

• Increased runoff and erosion • Increased runoff and erosion • Increased temporal variation
• Reduced infiltration • Reduced infiltration in streamflow
P1: GCE
(Table 3). Nutrient losses and greenhouse gas emissions may persist in cases of
repeated burning of pastures. Substantial hydrologic changes occur via decreased
plant transpiration, decreased soil infiltration, and increase variability in runoff and
stream flow (Table 3). In humid tropical systems, the global dominance of nutrient-
poor ultisol and oxisol soils (Figure 4c, Table 2) are widely implicated in the decline
of productivity in ranching systems (53, 109). Tropical deforestation has been a
regional syndrome largely driven by nutrient-poor soils, underlying a social and
political demand for cattle production (150). Climatological impacts of deforesta-
tion have been heavily studied, suggesting overall decreases in continental-scale
precipitation but possible increases in rainfall at the landscape-to-regional levels.

As the human population grows and land scarcity increases, intensification and
shifts in traditional animal production are occurring globally. Traditional extensive
pastoral systems are declining in some arid regions, and integrated pastoral farm
management is disappearing in tropical highlands. These are being replaced by

more concentrated grazing systems that can lead to greater degradation of pasture
land (151). In the northeastern United States, northwestern Europe, and densely
populated areas of Asia, animal production has become mechanized and dependent
on external fertilizer and feed inputs. Industrial meat production is growing rapidly.
From 1991 to 1993, it provided 37% of global meat production and 43% by 1996
(152). In these systems, animal waste exceeds the absorptive capacity of the land
and pollutes the surrounding environment (153). Excess nitrogen and phosphorus
from livestock is a substantial source of nonpoint pollution in the United States,
causing eutrophication of freshwater and marine ecosystems, toxic algal blooms,
and fish kills (154). Excess nitrogen in the environment can have many negative
impacts on human health (155). Other biogeochemical impacts of intensification
are only starting to be studied and quantified.
Knowledge Gaps and Research Needs

We contend that a combination of three regional syndromes—desertification,
woody encroachment, and deforestation—represents a major component of global
environmental change promoted and mediated by managed grazing activities.
There is a rich literature on the ways that cropland expansion and intensifica-
tion have altered both terrestrial and aquatic ecosystems (156, 157). There are also
clear effects of urban and suburban land use on regional ecological dynamics, and
the role of urbanization on climate is now being recognized at the global scale
(158). Managed grazing systems cover more of the Earth’s surface than any other
form of land use, yet pastoral operations are spatially diffuse and natural looking
(except in the case of deforestation). Nonetheless, managed grazing has resulted
in typological responses of ecosystems during the past few centuries if not be-
forehand. These responses can be organized by climate-edaphic conditions on a
regional basis, and in doing so, the global footprint of grazing systems appears to
be quite large. This synthesis helps stitch together this globally relevant land use
into a single framework for further study and perspective.
290 ASNER ET AL.

Figure 11 The grazing intensity at which ecosystems become degraded or

are significantly altered changes with bioclimatic setting, with greater impacts
and lower sustainability in very dry and very humid ecosystems.
As the result of the emergent themes from our synthesis, we hypothesize that
ecosystem responses to grazing vary along a bioclimatic gradient from arid to
mesic to humid environments (Figure 11). The literature suggests that mesic en-
vironments, such as in temperate grasslands, can endure the highest grazing in-
tensities (19). At some point, however, changes in the structure, biogeochem-
istry, and hydrology of these systems occur, and the degree of such changes
will be mediated by management practices. Arid environments are predisposed
to changes in ecosystem structure, biogeochemistry, and hydrology in ways that
truly degrade the land, even at relatively low grazing pressures. Humid forest
ecosystems (especially tropical) are immediately and significantly altered by graz-
ing systems, and degradation of pastureland often occurs under poor edaphic
conditions.
These hypotheses are well supported by the scientific literature, but our over-
all understanding of ecosystem responses to managed grazing remains somewhat
diffuse and fragmented for several reasons. The extent and intensity of managed
grazing operations are very poorly known at regional, continental, or global scales.
The GIS maps presented and synthesized here are state of the art, yet we recog-
nize the generality and inaccuracy of them, especially in regions, such as the
Indian subcontinent, the U.S. southwest, South America, and northern Mexico.
Remote sensing technologies cannot identify grazing lands without going to very
high-spatial and -biogeophysical resolution (37, 44, 80). Such approaches are im-
practical at the global scale, but a strategic global sampling would be tractable if we
knew how to stratify such a sampling, using process-level knowledge of grazing
systems and the resulting regional syndromes.
Our lack of process-based knowledge not only impedes our ability to make the
appropriate observations for analysis and monitoring, but it also limits our predic-
tive capabilities. We are currently unable to forecast the onset of desertification,
woody encroachment, or even deforestation because we lack the approaches to
understand the interaction between ecological, climatological, and socioeconomic
factors. Recent synthetic work has made some progress in this regarding deser-
tification (159) and deforestation (6), but the basic observations of when, where,
and under what conditions these regional syndromes occur are still lacking. Even
remote sensing studies of tropical deforestation produce wide-ranging estimates
of the amount of pastureland emplaced annually (5, 12). In sum, the observations
are limited by technological barriers, whereas the scientific understanding of graz-
ing systems and global change are limited by insufficient observations. Research
and progress are needed on both fronts to better understand the role of managed
grazing in the global environment.
ACKNOWLEDGMENTS
We thank A. Cooper and A. Warner for technical assistance, and R. Naylor and
P. Matson for thoughtful comments on the manuscript. This work was funded
by NASA New Investigator Program grant NAG5-8709, the National Science
Foundation, and the Mellon Foundation.
The Annual Review of Environment and Resources is online at

http://environ.annualreviews.org
LITERATURE CITED
1. Goldewijk K, Battjes CGM, Battjes Hurtt GC, Moorcroft PR, Birdsey RA.
JJ. 1997. A hundred year (1890–1990) 2000. Contributions of land-use history to
database for integrated environmental as- carbon accumulation in US forests. Sci-
sessments (HYDE, version 1.1). Rep. ence 290:1148–51
422514002, Natl. Inst. Public Health En- 5. Achard F, Eva HD, Stibig HJ, Mayaux P,
viron. (RIVM), Bilthoven, Neth. Gallego J, et al. 2002. Determination of
2. World Resour. Inst. 1990. World Re- deforestation rates of the world’s humid
sources 1990–1991. New York: Oxford tropical forests. Science 297:999–1002
Univ. Press 6. Geist HJ, Lambin EF. 2001. What drives
3. UN Environ. Programme. 1994. UN Earth tropical deforestation? LUCC Rep. Ser. 4,
summit. Convention on desertification. Land Use Cover Change Int. Project Off.,
Presented at UN Conf. Environ. Dev., Rio Louvain-la-Neuve, Belg.
de Janeiro, Brazil 7. Ramankutty N, Foley JA. 1998. Charac-
4. Caspersen JP, Pacala SW, Jenkins JC, terizing patterns of global land use: an
292 ASNER ET AL.
analysis of global croplands area. Glob. systems: What is the evidence? See Ref.
Biogeochem. Cycles 12:667–85 163, pp. 111–34
8. Hearn P, Hare T, Scruben P, Sherrill D, 18. Archer S, Schimel DS, Holland EA. 1995.
LaMar C, Tsushima P. 2001. Global GIS Mechanisms of shrubland expansion: land
database: digital atlas of the world. Rep. use, climate or CO2. Clim. Chang. 29:91–
Digit. Data Ser. DDS-62-H, US Geol. 99
Surv., Reston, VA 19. Milchunas DG, Lauenroth WK. 1993.
9. Field CB, Randerson JT, Malmström CM. Quantitative effects of grazing on vege-
1995. Global net primary production: tation and soils over a global range of en-
combining ecology and remote sensing. vironments. Ecol. Monogr. 63:327–66
Remote Sens. Environ. 51:74–88 20. Valone TJ, Meyer M, Brown JH, Chew
10. US Dep. Agric. 1999. Soil taxonomy: a RM. 2002. Timescale of perennial grass
basic system of soil classification for mak- recovery in desertified arid grasslands fol-
ing and interpreting soil surveys. Rep. lowing livestock removal. Conserv. Biol.
486, USDA, Nat. Resour. Conserv. Serv. 16:995–1002

Washington, DC 21. Hughes L. 2003. Climate change and Aus-
11. McGrath DA, Smith CK, Gholz HL, tralia: trends, projections and impacts.
Oliveira F. 2001. Effects of land-use Austral Ecol. 28:423–43
change on soil nutrient dynamics in 22. Kerley GIH, Knight MH, DeKock
Amazônia. Ecosystems 4:625–45 M. 1995. Desertification of subtropical
12. DeFries RS, Houghton RA, Hansen MC, thicket in the Eastern Cape, South Africa:
Field CB, Skole D, Townshend J. 2002. Are there alternatives? Environ. Monit.
Carbon emissions from tropical deforesta- Assess. 37:211–30
tion and regrowth based on satellite ob- 23. Saleem MAM. 1998. Nutrient balance
servations for the 1980s and 1990s. Proc. patterns in African livestock systems.
Natl. Acad. Sci. USA 99:14256–61 Agric. Ecosyst. Environ. 71:241–54
13. Rosegrant MW, Paisner MS, Meijer S, 24. Oba G, Weladji RB, Lusigi WJ, Stenseth
Witcover J. 2001. Global Food Projec- NC. 2003. Scale-dependent effects of
tions to 2020: Emerging Trends and Al- grazing on rangeland degradation in
ternative Futures. New York: Int. Food northern Kenya: a test of equilibrium and
Policy Research Inst. non-equilibrium hypotheses. Land De-
14. Prince SD, Colstoun EBd, Kravitz LL. grad. Dev. 14:83–94
1998. Evidence from rain-use efficiencies 25. Oconnor TG. 1993. The influence of rain-
does not indicate extensive Sahelian de- fall and grazing on the demography of
sertification. Glob. Change Biol. 4:359– some African savanna grasses—a ma-
74 trix modeling approach. J. Appl. Ecol.
15. Schlesinger WH, Reynolds JF, Cunning- 30:119–32
ham GL, Huenneke LF, Jarrell WM, et al. 26. D’Antonio CM, Vitousek PM. 1992. Bi-
1990. Biological feedbacks in global de- ological invasions by exotic grasses, the
sertification. Science 247:1043–48 grass/fire cycle, and global change. Annu.
16. Okin GS, Murray B, Schlesinger WH. Rev. Ecol. Syst. 23:63–87
2001. Degradation of sandy arid shrub- 27. Buffington LC, Herbel CH. 1965. Vege-
land environments: observations, process tational changes on a sediment grassland
modelling, and management implications. range from 1858 to 1963. Ecol. Monogr.
J. Arid Environ. 47:123–44 35:139–64
17. Ash AJ, Stafford-Smith DM, Abel N. 28. Mabbutt JA. 1984. A new global assess-
2002. Land degradation and secondary ment of the status and trends of desertifi-
production in semi-arid and arid grazing cation. Environ. Conserv. 11:103–13
29. Milton SJ, Dean WRJ. 1995. South New Mexico: II. Field plots. Biogeochem-
Africa’s arid and semiarid rangelands: istry 49:69–86
Why are they changing and can they be 40. Hartley AE, Schlesinger WH. 2000. Envi-
restored? Environ. Monit. Assess. 37:245– ronmental controls on nitric oxide emis-
64 sions from northern Chihuahuan desert
30. Aagesen D. 2000. Crisis and conservation soils. Biogeochemistry 50:279–300
at the end of the world: sheep ranching 41. Schlesinger WH, Peterjohn WT. 1991.
in Argentine Patagonia. Environ. Conserv. Processes controlling ammonia volatiliza-
27:208–15 tion from Chihuahuan desert soils. Soil
31. van de Koppel J, Rietkerk M, van Biol. Biochem. 23:637–42
Langevelde F, Kumar L, Klausmeier CA, 42. Peterjohn WT, Schlesinger WH. 1991.
et al. 2002. Spatial heterogeneity and ir- Factors controlling denitrification in a
reversible vegetation change in semiarid Chihuahuan desert ecosystem. Soil Sci.
grazing systems. Am. Nat. 159:209–18 Soc. Am. J. 55:1694–701
32. Gao Q, Reynolds JF. 2003. Historical 43. Huenneke LF, Anderson JP, Remmenga
shrub-grass transitions in the northern M, Schlesinger WH. 2002. Desertifica-
Chihuahuan Desert: modeling the effects tion alters patterns of aboveground net pri-
of shifting rainfall seasonality and event mary production in Chihuahuan ecosys-
size over a landscape gradient. Glob. tems. Glob. Change Biol. 8:247–64
Change Biol. 9:1475–93 44. Asner GP, Archer S, Hughes RF, Ans-
33. Parsons JJ. 1970. Spread of African pas- ley RJ, Wessman CA. 2003. Net changes
ture grasses in the American tropics. J. in regional woody vegetation cover
Range Manag. 25:12–17 and carbon storage in Texas drylands,
34. Hughes F, Vitousek PM. 1993. Barriers 1937–1999. Glob. Change Biol. 9:316–
to shrub reestablishment following fire in 35
the seasonal submontane zone of Hawaii. 45. Jackson RB, Banner JL, Jobbagy EG,
Oecologia 93:557–63 Pockman WT, Wall DH. 2002. Ecosystem
35. Asner GP, Beatty SW. 1996. Effects of carbon loss with woody plant invasion of
an African grass invasion on Hawai- grasslands. Nature 418:623–26
ian shrubland nitrogen biogeochemistry. 46. Hughes RF, Archer S, Asner GP, Wess-
Plant Soil 186:205–11 man CA, McMurtrie RE, et al. 2004.
36. Mack MC, D’Antonio CM, Ley RE. 2001. Changes in primary production and car-
Alteration of ecosystem nitrogen dynam- bon and nitrogen pools accompanying
ics by exotic plants: a case study of C-4 woody encroachment in a north Texas sa-
grasses in Hawaii. Ecol. Appl. 11:1323– vanna. Ecosystems In press
35 47. Hibbard KA, Archer S, Schimel DS,
37. Asner GP, Borghi CE, Ojeda RA. Valentine DW. 2001. Biogeochemical
2003. Desertification in central Argentina: changes accompanying woody plant en-
Changes in ecosystem carbon and ni- croachment in a subtropical savanna.
trogen from imaging spectroscopy. Ecol. Ecology 82:1999–2011
Appl. 13:629–48 48. Geesing D, Felker P, Bingham RL. 2000.
38. Gallardo A, Schlesinger WH. 1992. Car- Influence of mesquite (Prosopis glandu-
bon and nitrogen limitations of soil micro- losa) on soil nitrogen and carbon devel-
bial biomass in desert ecosystems. Bio- opment: Implications for global carbon
geochemistry 18:1–17 sequestration. J. Arid Environ. 46:157–
39. Schlesinger WH, Ward TJ, Anderson J. 80
2000. Nutrient losses in runoff from grass- 49. Martin RE, Asner GP. 2004. Regional
land and shrubland habitats in southern nitric oxide emissions following woody
294 ASNER ET AL.
encroachment: linking imaging spec- global change. Biol. Fertil. Soils 27:221–
troscopy and field studies. Ecosystems. In 29
press 61. Robertson G. 1989. Group report, Trace
50. Martin RE, Asner GP, Ansley RJ, Mosier gas exchange and the chemical and phys-
AR. 2003. Effects of woody encroach- ical climate: critical interactions. See
ment on soil nitrogen oxide emissions in Ref. 162, pp. 24–36
a temperate savanna. Ecol. Appl. 13:897– 62. Firestone MK, Davidson EA. 1989. Mi-
910 crobiological basis of NO and N2O pro-
51. Markewitz D, Davidson E, Moutinho P, duction and consumption in soil. See
Nepstad D. 2004. Nutrient loss and redis- Ref. 162, pp. 7–21
tribution after forest clearing on a highly 63. Hartley AE. 1997. Environmental controls
weathered soil in Amazonia. Ecol. Appl. on nitrogen cycling in northern Chihua-
In press han desert soils. PhD thesis. Duke Univ.,
52. Guild LS, Kauffman JB, Ellingson LJ, Durham, NC
Cummings DL, Castro EA. 1998. Dy- 64. Mosier A, Schimel D, Valentine D, Bron-
namics associated with total aboveground son K, Parton W. 1991. Methane and ni-
biomass, C, nutrient pools, and biomass trous oxide fluxes in native, fertilized and
burning of primary forest and pasture in cultivated grasslands. Nature 350:330–32
Rondonia, Brazil during SCAR-B. J. Geo- 65. Bouwman AF, Lee DS, Asman WAH,
phys. Res. Atmos. 103:32091–100 Dentner FJ, Van Der Hoek KW, Olivier
53. Fearnside PM, Barbosa RI. 1998. Soil car- JGJ. 1997. A global high-resolution emis-
bon changes from conversion of forest to sion inventory for ammonia. Glob. Bio-
pasture in Brazilian Amazonia. For. Ecol. geochem. Cycles 11:561–87
Manag. 108:147–66 66. Yamulki S, Jarvis SC, Owen P. 1998. Ni-
54. Schlesinger WH, Pilmanis AM. 1998. trous oxide emissions from excreta ap-
Plant-soil interactions in deserts. Biogeo- plied in a simulated grazing pattern. Soil
chemistry 42:169–87 Biol. Biochem. 30:491–500
55. Schlesinger WH, Abrahams AD, Parsons 67. Johnson DE, Johnson KA, Ward GM, Bra-
AJ, Wainwright J. 1999. Nutrient losses in nine ME. 2000. Ruminant and other ani-
runoff from grassland and shrubland habi- mals. In Atmospheric Methane: Its Role in
tats in southern New Mexico: I. Rainfall the Global Environment, ed. MAK Kahill,
simulation experiments. Biogeochemistry pp. 112–33. Berlin: Springer-Verlag
45:21–34 68. Stephenson NL. 1990. Climatic control of
56. Branson FA, Gifford GF, Renard KG, vegetation distrbution: the role of the wa-
Hadley RF. 1981. Rangeland Hydrology. ter balance. Am. Nat. 135:649–70
Dubuque, IA: Kendall/Hunt. 339 pp. 69. Parton WJ, Lauenroth WK, Smith FM.
57. Sparrow AD, Friedel MH, Tongway DJ. 1981. Water loss from a shortgrass steppe.
2003. Degradation and recovery pro- Agric. Meteorol. 24:97–109
cesses in arid grazing lands of central Aus- 70. Frank DA, Inouye RS. 1994. Temporal
tralia. Part 3: Implications at landscape variation in actual evapotranspiration of
scale. J. Arid Environ. 55:349–60 terrestrial ecosystems—patterns and eco-
58. Belnap J. 1995. Surface disturbances: logical implications. J. Biogeogr. 21:401–
their role in accelerating desertification. 11
Environ. Monit. Assess. 37:39–57 71. Bremer DJ, Auen LM, Ham JM, Owensby
59. Intergov. Panel Clim. Change. 1995. The CE. 2001. Evapotranspiration in a prairie
Science of Climate Change. Cambridge: ecosystem: effects of grazing by cattle.
IPCC Working Group I Agron. J. 93:338–48
60. Mosier AR. 1998. Soil processes and 72. Svejcar T, Christiansen S. 1987. Grazing
effects on water relations of Caucasian 83. Dirmeyer PA, Shukla J. 1994. Albedo as a
bluestem. J. Range Manag. 40:15–18 modulator of climate response to tropical
73. Wraith JM, Johnson DA, Hanks RJ, Sis- deforestation. J. Geophys. Res. 99:20863–
son DV. 1987. Soil and plant water re- 77
lations in a crested wheatgrass pasture: 84. Bryant AD, Johnson LF, Brazel AJ,
response to spring grazing by cattle. Oe- Balling RC, Hutchinson CF, Beck LR.
cologia 73:573–78 1990. Measuring the effect of over-
74. Naeth MA, Chanasyk DS. 1995. Grazing grazing in the Sonoran Desert. Clim.
effects on soil-water in Alberta foothills Chang. 17:243–64
fescue grasslands. J. Range Manag. 48: 85. Michalek JL, Colwell JE, Roller NEG,
528–34 Miller NA, Kasischke ES, Schlesinger

75. Naeth MA, Chanasyk DS, Rothwell RL, WH. 2001. Satellite measurements of
Bailey AW. 1991. Grazing impacts on albedo and radiant temperature from
soil-water in mixed prairie and fescue semi-desert grassland along the Arizona/
grassland ecosystems of Alberta. Can. J. Sonora border. Clim. Chang. 48:417–25

Soil Sci. 71:313–25 86. Archer S. 1994. Woody plant encroach-
76. Trimble SW, Mendel AC. 1995. The cow ment into southwestern grasslands and
as a geomorphic agent—a critical review. savannas: rates, patterns, and proximate
Geomorphology 13:233–53 causes. In Ecological Implications of
77. Marshall JK. 1973. Drought, land use Livestock Herbivory in the West, ed. M
and soil erosion. In The Environmen- Vavra, WA Laycock, RD Pieper, pp. 13–
tal, Economic, and Social Significance of 68. Denver, CO: Soc. Range Manag.
Drought., ed. JV Lovett, pp. 55–77. Syd- 87. van Auken WO. 2000. Shrub invasions
ney, Aust.: Angus & Robertson of North American semi-arid grasslands.
78. Hendricks BA. 1942. Effect of grass lit- Annu. Rev. Ecol. Syst. 31:197–216
ter on infiltration of rainfall on granitic 88. Archer S, Scifres C, Bassham CR. 1988.
soils in a semidesert shrub grass area. Rep. Autogenic succession in a subtropical sa-
Note 96, USDA, For. Serv., Southwest vanna: conversion of grassland to thorn
For. Range Exp. Station, Albuquerque, woodland. Ecol. Monogr. 58:111–27
NM 89. Brown JR, Archer S. 1989. Woody plant
79. Fiedler FR, Frasier GW, Ramirez JA, invasion of grasslands: establishment of
Ahuja LR. 2002. Hydrologic response honey mesquite (Prosopis glandulosa var.
of grasslands: effects of grazing, interac- glandulosa) on sites differing in herba-
tive infiltration, and scale. J. Hydrol. Eng. ceous biomass and grazing history. Oe-
7:293–301 cologia 80:19–26
80. Pickup G, Bastin GN, Chewings VH. 90. Hughes RF, Archer S, Asner GP, Wess-
1994. Remote-sensing-based condition man CA. 1999. Ecosystem-scale impli-
assessment for nonequilibrium range- cations of woody encroachment: storage
lands under large-scale commercial graz- and production of mesquite woodlands
ing. Ecol. Appl. 4:497–517 and their impact on C and N dynamics of
81. HilleRisLambers R, Rietkerk M, van den mixed prairie ecosystems in North Texas.
Bosch F, Prins HHT, de Kroon H. 2001. Presented at Ecol. Soc. Am. Annu. Meet.,
Vegetation pattern formation in semi-arid Tucson, AZ
grazing systems. Ecology 82:50–61 91. Asner GP, Wessman CA, Archer S. 1998.
82. Charney J, Stone PH, Quirk WJ. 1975. Scale dependence of absorption of photo-
Drought in the Sahara: a biogeophysical synthetically active radiation in terrestrial
feedback mechanism. Science 187:434– ecosystems. Ecol. Appl. 8:1003–21
35 92. Tobler MW, Cochard R, Edwards PJ.
296 ASNER ET AL.
2003. The impact of cattle ranching tional Types: Their Relevance to Ecosys-
on large-scale vegetation patterns in a tem Properties and Global Change, ed.
coastal savanna in Tanzania. J. Appl. Ecol. TM Smith, HH Shugart, FI Woodward,
40:430–44 pp. 217–33. New York: Cambridge Univ.
93. Harris AT, Asner GP, Miller ME. 2003. Press
Changes in vegetation structure af- 102. Breshears DD, Barnes FJ. 1999. Inter-
ter long-term grazing in pinyon-juniper relationships between plant functional
ecosystems: integrating imaging spec- types and soil moisture heterogeneity
troscopy and field studies. Ecosystems 6: for semiarid landscapes within the grass-
368–83 land/forest continuum: a unified con-
94. Mapfumo E, Chanasyk DS, Naeth MA, ceptual model. Landsc. Ecol. 14:465–
Baron VS. 1999. Soil compaction under 78
grazing of annual and perennial forages. 103. Twerdoff DA, Chanasyk DS, Naeth MA,
Can. J. Soil Sci. 79:191–99 Baron VS, Mapfumo E. 1999. Soil wa-
95. Friedel MH, Sparrow AD, Kinloch JE, ter regimes of rotationally grazed peren-
Tongway DJ. 2003. Degradation and re- nial and annual forages. Can. J. Soil Sci.
covery processes in and grazing lands of 79:627–37
central Australia. Part 2: Vegetation. J. 104. Breshears DD, Nyhan JW, Heil CE,
Arid Environ. 55:327–48 Wilcox BP. 1998. Effects of woody plants
96. Cingolani AM, Cabido MR, Renison D, on microclimate in a semiarid wood-
Solis VN. 2003. Combined effects of envi- land: soil temperature and evaporation in
ronment and grazing on vegetation struc- canopy and intercanopy patches. Int. J.
ture in Argentine granite grasslands. J. Plant Sci. 159:1010–17
Veg. Sci. 14:223–32 105. Breshears DD, Rich PM, Barnes FJ,
97. Fuhlendorf SD, Briske DD, Smeins FE. Campbell K. 1997. Overstory-imposed
2001. Herbaceous vegetation change in heterogeneity in solar radiation and soil
variable rangeland environments: the rel- moisture in a semiarid woodland. Ecol.
ative contribution of grazing and climatic Appl. 7:1201–16
variability. Appl. Veg. Sci. 4:177–88 106. Grover HD, Musik HB. 1990. Shrubland
98. Pettit NE, Froend RH. 2001. Long-term encroachment in southern New Mexico,
changes in the vegetation after the ces- U.S.A.: an analysis of desertification pro-
sation of livestock grazing in Eucalyptus cesses in the American Southwest. Cli-
marginata (jarrah) woodland remnants. mat. Chang. 17:305–30
Austral Ecol. 26:22–31 107. Hoffman WA, Jackson RB. 2000.
99. Boutton TW, Archer SR, Midwood AJ, Vegetation-climate feedbacks in the con-
Zitzer SF, Bol R. 1998. Delta C-13 values version of tropical savanna to grassland.
of soil organic carbon and their use in doc- J. Clim. 13:1593–602
umenting vegetation change in a subtrop- 108. Weaver CP, Avissar R. 2001. Atmospheric
ical savanna ecosystem. Geoderma 82:5– disturbances caused by human modifica-
41 tion of the landscape. Bull. Am. Meteorol.
100. Walker H. 1971. Natural savannahs as a Soc. 82:269–81
transition to the arid zone. In Ecology of 109. Uhl C, Buschbacher R, Serrão EAS. 1988.
Tropical and Subtropical Vegetation, ed. Abandoned pastures in eastern Amazonia.
JH Burnett, pp. 238–65. New York: Van I. Patterns of plant succession. J. Ecol.
Nostrand Reinhold 76:663–81
101. Sala OE, Lauenroth WK, Golluscio RA. 110. Asner GP, Bustamante MMC, Townsend
1997. Plant functional types in temper- AR. 2003. Scale dependence of biophysi-
ate semi-arid regions. In Plant Func- cal structure in deforested areas bordering
the Tapajos National Forest, Central Ama- 1998. Fire in the Brazilian Amazon 2.
zon. Remote Sens. Environ. 87:507–20 Biomass, nutrient pools and losses in cat-
111. Kauffman JB, Sanford RL Jr., Cummings tle pastures. Oecologia 113:415–27
DL, Salcedo IH, Sampaio EVSB. 1993. 121. Davidson EA, de Carvalho CJR, Vieira
Biomass and nutrient dynamics associ- ICG, Figueiredo RO, Moutinho P, et al.
ated with slash fires in neotropical dry 2004. Nutrient limitation of biomass
forests. Ecology 74:140–51 growth in a tropical secondary forest:
112. Gimeno-Garcia E, Andreu V, Rubio J. early results of a nitrogen and phospho-
2000. Changes in organic matter, nitro- rus amendment experiment. Ecol. Appl.
gen, phosphorus, and cations in soil as In press
a result of fire and water erosion in a 122. Guggenberger G, Haumaier L, Thomas

Mediterranean landscape. Eur. J. Soil Sci. RJ, Zech W. 1996. Assessing the organic
51:201–10 phosphorus status of an oxisol under trop-
113. Belillus CM, Roda F. 1993. The effects ical pastures following native savanna us-
of fire on water-quality, dissolved nutrient ing 31P NMR spectroscopy. Biol. Fertil.
losses and the export of particulate matter Soils 23:332–39
from dry heathland catchments. J. Hydrol. 123. Garcia-Montiel DC, Neill C, Melillo J,
150:1–17 Thomas S, Steudler PA, Cerri CC. 2000.
114. Ellingson LJ, Kauffman JB, Cummings Soil phosphorus transformations follow-
DL, Sanford RL Jr., Jaramillo VJ. 2000. ing forest clearing for pasture in the
Soil N dynamics associated with defor- Brazilian Amazon. Soil Sci. Soc. Am. J.
estation, biomass burning, and pasture 64:1792–804
conversion in a Mexican tropical dry for- 124. Townsend AR, Asner GP, Cleveland CC,
est. For. Ecol. Manag. 137:41–51 Lefer ME, Bustamante MMC. 2002. Un-
115. Emmerich WE. 1999. Nutrient dynamics expected changes in soil phosphorus dy-
of rangeland burns in southeast Arizona. namics along pasture chronosequences
J. Range Manag. 52:606–14 in the humid tropics. J. Geophys. Res.
116. Sanchez PA. 1976. Properties and Man- 107(D20):8067–76
agement of Soils in the Tropics. New York: 125. Asner GP, Townsend AR, Bustamante
John Wiley MMC. 1999. Spectrometry of pasture
117. Vitousek PM, Chadwick OA, Crews TE, condition and biogeochemistry in the cen-
Fownes JH, Hendricks DM, Herbert D. tral Amazon. Geophys. Res. Lett. 26:
1997. Soil and ecosystem development 2769–72
across the Hawaiian Islands. GSA Today 126. Gehring C, Denich M, Kanashiro M, Vlek
7:1–8 PLG. 1999. Response of secondary veg-
118. Fernandes ECM, Biot Y, Castilla C, Canto etation in eastern Amazonia to relaxed
AC, Matos JC, et al. 1997. The impact of nutrient availability constraints. Biogeo-
selective logging and forest conservation chemistry 45:223–41
for subsistence agriculture and pastures on 127. Davidson EA, Kingerlee W. 1997. A
terrestrial nutrient dynamics in the Ama- global inventory of nitric oxide emissions
zon. Ciencia e Cultura: J. Braz. Assoc. from soils. Nutr. Cycl. Agroecosyst. 48:
Adv. Sci. 49:37–47 37–50
119. Murty D, Kirschbaum MUF, McMurtrie 128. Keller M, Reiners WA. 1994. Soil-
RE, McGilvray A. 2002. Does conversion atmosphere exchange of nitrous oxide, ni-
of forest to agricultural land change soil tric oxide, and methane under secondary
carbon and nitrogen? A review of the lit- succession of pasture to forest in the At-
erature. Glob. Change Biol. 8:105–23 lantic lowlands of Costa Rica. Glob. Bio-
120. Kauffman JB, Cummings DL, Ward DE. geochem. Cycles 8:399–409
298 ASNER ET AL.
129. Davidson EA, Keller MK, Erickson HE, 140. Costa MH, Botta A, Cardille JA. 2003.
Verchot LV, Veldkamp E. 2000. Test- Effects of large-scale changes in land
ing a conceptual model of soil emissions cover on the discharge of the Tocantins
of nitrous and nitric oxides. BioScience River, southeastern Amazonia. J. Hydrol.
50:667–80 283:206–17
130. Crutzen PJ, Andreae MO. 1990. Biomass 141. Calder I. 1999. The Blue Revolution, Land
burning in the tropics: impact on atmo- Use and Integrated Water Resources Man-
spheric chemistry and biogeochemical cy- agement. London, UK: Earthscan
cles. Science 250:1669–78 142. Bruijnzeel LA. 2001. Forest hydrol-
131. Reich PB, Peterson DW, Wedin DA, ogy. In The Forests Handbook, ed. JC
Wrage K. 2001. Fire and vegetation ef- Evans, pp. 56–78. Oxford, UK: Blackwell
fects on productivity and nitrogen cy- Scientific
cling across a forest-grassland continuum. 143. Nobre CA, Sellers PJ, Shukla J.
Ecology 82:1703–19 1991. Amazon deforestation and re-
132. Scholes M, Andreae MO. 2000. Biogenic gional climate change. J. Clim. 4:957–
and pyrogenic emissions from Africa and 88
their impact on the global atmosphere. 144. Silva Dias MAF, Rutledge S, Kabat P,
Ambio 29:23–29 Silva Dias PL, Nobre C, et al. 2002.
133. Thompson AM, Witte JC, Hudson RD, Cloud and rain processes in a biosphere-
Guo H, Herman JR, Fujiwara M. 2001. atmosphere interaction context in the
Tropical tropospheric ozone and biomass Amazon Region. J. Geophy. Res. 107:
burning. Science 291:2128–32 8072–92
134. Artaxo P, Martins JV, Yamasoe MA, 145. Werth D, Avissar R. 2002. The local and
Procopio AS, Pauliquevis TM, et al. global effects of Amazon deforestation. J.
2002. Physical and chemical properties Geophys. Res. 107:8087–95
of aerosols in the wet and dry seasons 146. Gedney N, Valdes PJ. 2000. The effect of
in Rondonia, Amazonia. J. Geophys. Res. Amazonian deforestation on the Northern
107:8081–95 Hemisphere circulation and climate. Geo-
135. Andreae MO, Crutzen PJ. 1997. Atmo- phys. Res. Lett. 27:2052–56
spheric aerosols: biogeochemical sources 147. Bahre CJ. 1995. Human impacts on the
and roles in atmospheric chemistry. Sci- grasslands of southeastern Arizona. In
ence 276:1052–57 The Desert Grassland, ed. MP McClaran,
136. Hobbs PT, Reid JS, Kotchenruther RA, TR Van Devender, pp. 230–64. Tucson,
Ferek RJ, Weiss R. 1997. Direct radiative AZ: Univ. Ariz. Press
forcing by smoke from biomass burning. 148. Holmgren M, Scheffer M. 2001. El Niño
Science 275:1776–78 as a window of opportunity for the restora-
137. Giambelluca TW. 2002. Hydrology of tion of degraded arid ecosystems. Ecosys-
altered tropical forest. Hydrol. Process. tems 4:151–59
16:1665–69 149. Bachelet D, Lenihan JM, Daly C, Neil-
138. Moreira MZ, Sternberg LDL, Martinelli son RP. 2000. Interactions between fire,
LA, Victoria RL, Barbosa EM, et al. 1997. grazing and climate change at Wind Cave
Contribution of transpiration to forest am- National Park, SD. Ecol. Model. 134:229–
bient vapour based on isotopic measure- 44
ments. Glob. Change Biol. 3:439–50 150. Hecht S, Cockburn A. 1990. The Fate of
139. Bruijnzeel LA. 1990. Hydrology of Moist the Forest. New York: Harper Perennial.
Forests and the Effects of Conversion: A 357 pp.
State of Knowledge Review. Amsterdam, 151. de Haan C, Steinfeld H, Blackburn H.
Neth.: Free Univ. 224 pp. 1997. Livestock and the environment. Rep.
UN Food Agric. Organ. Fressingfield, nomic, and economic aspects of fertilizer

Neth. management. Science 280:112–14
152. Steinfeld H, de Haan C, Blackburn H. 158. Goldreich Y. 1995. Climate studies in
1997. Options to address livestock- Israel—a review. Atmos. Environ. 29:
environment interactions. World Animal 467–78
Rev. 88:15–20 159. Stafford Smith DM, Reynolds JF. 2002.
153. UN Food Agric. Organ. 2002. http:// The Dahlem desertification paradigm: a
www.agrifood-forum.net/practices/sector/ new approach to an old problem. See Ref.
livestock/21systems.asp 163, pp. 403–24
154. Carpenter SR, Caraco NF, Correll DL, 160. Ramankutty N, Foley JA. 1999. Esti-
Howarth RW, Sharpley AN, Smith VH. mating historical changes in global land
1998. Nonpoint pollution of surface wa- cover: croplands from 1700 to 1992. Glob.
ters with phosphorus and nitrogen. Ecol. Biogeochem. Cycles 13:997–1027
Appl. 8:559–68 161. Los SO, Collatz GJ, Sellers PJ, Malm-
155. Townsend AR, Howarth RW, Bazzaz FA, ström CM, Pollack NH, et al. 2000. A
Booth MS, Cleveland CC, et al. 2003. Hu- global 9-year biophysical land surface
man health effects of a changing global ni- dataset from NOAA AVHRR data. J. Hy-
trogen cycle. Front. Ecol. Environ. 1:240– drometeorol. 1:183–99
46 162. Andreae MO, Schimel DS, eds. 1989.
156. Vitousek PM, Aber JD, Howarth RW, Exchange of Trace Gases between Ter-
Likens GE, Matson PA, et al. 1997. Hu- restrial Ecosystems and the Atmosphere.
man alteration of the global nitrogen Chichester, UK: John Wiley
cycle: sources and consequences. Ecol. 163. Reynolds JF, Stafford Smith DM, eds.
Appl. 7:737–50 2002. Global Desertification: Do Humans
157. Matson PA, Naylor R, Ortiz-Monasterio I. Cause Deserts? Berlin: Dahlem Univ.
1998. Integration of environmental, agro- Press
HI-RES-EG29-11-Asner.qxd
10/16/04
11:39 AM
Page 1
Figure 1 The present global distribution of grazing extent and intensity (stocking rates), derived by combining References 1 and 2.
GRAZING SYSTEMS AND GLOBAL CHANGE
C-1
C-2
ASNER ET AL.
10/16/04
11:39 AM
Page 2
Figure 3 Global distribution of managed grazing systems overlaid on the global distribution of biomes (1, 160).
10/16/04
11:39 AM
Page 3
Figure 4a Global distribution of managed grazing systems and actual evapotranspiration (AET), a metric of bioclimatic stress. Derived
by combining References 1 and 8.
GRAZING SYSTEMS AND GLOBAL CHANGE
C-3
C-4
ASNER ET AL.
10/16/04
11:39 AM
Page 4
Figure 4c Global distribution of managed grazing systems and soil taxonomic order. Derived by combining References 1 and 10.
P1: JRX
September 29, 2004 21:55 Annual Reviews AR227-FM
Annual Review of Environment and Resources

Volume 29, 2004
CONTENTS
I. EARTH’S LIFE SUPPORT SYSTEMS

Current Uncertainties in Assessing Aerosol Effects on Climate,
Surabi Menon 1
Marine Reserves and Ocean Neighborhoods: The Spatial Scale of Marine
Populations and Their Management, Stephen R. Palumbi 31

Elemental Cycles: A Status Report on Human or Natural Dominance,
R.J. Klee and T.E. Graedel 69
II. HUMAN USE OF ENVIRONMENT AND RESOURCES
Prospects for Carbon Capture and Storage Technologies, Soren Anderson
and Richard Newell 109
Plant Genetic Resources for Food and Agriculture: Assessing Global
Availability, Cary Fowler and Toby Hodgkin 143
Construction Materials and the Environment, Arpad Horvath 181
Contested Terrain: Mining and the Environment, Gavin Bridge 205
Grazing Systems, Ecosystem Responses, and Global Change,
Gregory P. Asner, Andrew J. Elmore, Lydia P. Olander,
Roberta E. Martin, and A. Thomas Harris 261
III. MANAGEMENT AND HUMAN DIMENSIONS
Assessing the Costs of Electricity, Daniel M. Kammen and Sergio Pacca 301
Advances in Energy Forecasting Models Based on Engineering
Economics, Ernst Worrell, Stephan Ramesohl, and Gale Boyd 345
Energy Management and Global Health, Majid Ezzati, Robert Bailis,
Daniel M. Kammen, Tracey Holloway, Lynn Price, Luis A. Cifuentes,
Brendon Barnes, Akanksha Chaurey, and Kiran N. Dhanapala 383
Energy Infrastructure and Security, Alexander E. Farrell, Hisham Zerriffi,
and Hadi Dowlatabadi 421
INDEXES
Subject Index 471
Cumulative Index of Contributing Authors, Volumes 20–29 493
Cumulative Index of Chapter Titles, Volumes 20–29 497
ix
P1: JRX
September 29, 2004 21:55 Annual Reviews AR227-FM
x CONTENTS
ERRATA
An online log of corrections to Annual Review of
Environment and Resources chapters may be found
at http://environ.annualreviews.org

Asner Et Al 2004 Grazing Systems, Ecosystem Responses, and PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Asner Et Al 2004 Grazing Systems, Ecosystem Responses, and PDF

Uploaded by

Copyright:

Available Formats

16 Oct 2004 10:42 AR AR227-EG29-08.tex AR227-EG29-08.

sgm LaTeX2e(2002/01/18) P1: GCE

Annu. Rev. Environ. Resour. 2004. 29:261–99

GRAZING SYSTEMS, ECOSYSTEM RESPONSES, AND

Roberta E. Martin,1 and A. Thomas Harris1

Key Words agriculture, deforestation, desertification, land-use change, woody

262 ASNER ET AL.

Knowledge Gaps and Research Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289

GRAZING SYSTEMS AND GLOBAL CHANGE 263

the impacts of managed grazing on ecosystem structure, biogeochemistry, hydrol-

THE GLOBAL FOOTPRINT OF GRAZING SYSTEMS

GIS Analysis of Bioclimatic and Edaphic Conditions

GRAZING SYSTEMS AND GLOBAL CHANGE 265

area of global grazed biome grazing biome biome biome biome

Savanna 19.31 15 9.48 49.1 595 781 0.76 0.69 0.90

Tundra 7.32 6 0.17 2.3 431 247 1.75 1.74 0.01

GRAZING SYSTEMS AND GLOBAL CHANGE 267

Mean annual AET is a metric of the average bioclimatic conditions of an ecosys-

(monthly temporal resolution; 1◦ × 1◦ spatial resolution) to map the global inter-

GRAZING SYSTEMS AND GLOBAL CHANGE 269

Evergreen/deciduous mixed Inceptisols Gelisols 

Regional Syndromes in the Global Grazing Footprint

270 ASNER ET AL.

GRAZING SYSTEMS AND GLOBAL CHANGE 271

To present our synthetic perspective on the common syndromes inherent to

pronounced climatic variability (Figure 5) (3). Woody encroachment, as defined

272 ASNER ET AL.

of studies describe changes in ecosystem structure that appear to be caused or

soil moisture (see the Hydrology section).

Figure 6 Processes mediating desertification in arid grazing systems.

GRAZING SYSTEMS AND GLOBAL CHANGE 273

geographic expansion (26).

GRAZING SYSTEMS AND GLOBAL CHANGE 275

biological activity of the crusts (58).

276 ASNER ET AL.

GRAZING SYSTEMS AND GLOBAL CHANGE 277

Mendel (76) found that infiltration decreased from approximately 50 mm h−1 on

278 ASNER ET AL.

There are hundreds of documented cases of increased woody plant cover in

Figure 9 Processes mediating woody vegetation encroachment in semiarid and mesic

GRAZING SYSTEMS AND GLOBAL CHANGE 279

280 ASNER ET AL.

Figure 10 Commonly reported responses of five-dryland ecosystem structural properties

GRAZING SYSTEMS AND GLOBAL CHANGE 281

Woody encroachment has a measurable effect on soil nitrogen oxide emissions. In

282 ASNER ET AL.

tation events (100). Particularly in the presence of grazing, high-rainfall events or

GRAZING SYSTEMS AND GLOBAL CHANGE 283

though variation in pasture cover causes concomitant variation in biogeochemical,

the effects of this variation.

284 ASNER ET AL.

1900 kg N ha−1 is lost, which is equivalent to ∼90% of the aboveground pool in a

Trace-Gas and Aerosol Emissions

GRAZING SYSTEMS AND GLOBAL CHANGE 285

286 ASNER ET AL.

ECOSYSTEM RESPONSES TO MANAGED GRAZING

GRAZING SYSTEMS AND GLOBAL CHANGE 287

TABLE 3 Relative effects of land-cover change due to grazing on land surface

Deforestation Woody proliferation Desertification

Ecosystem properties Desertification Woody encroachment Deforestation

Vegetation structure • Decreased herbaceous, • Increased woody cover • Increased herbaceous

• Increased spatial heterogeneity

of soil moisture of soil moisture • Decreased infiltration